Lexical inhibition of neighbors during visual word recognition: an unmasked priming investigation.

brain research 1604 (2015) 35–51

Available online at www.sciencedirect.com

www.elsevier.com/locate/brainres

Research Report

Lexical inhibition of neighbors during visual word recognition: An unmasked priming investigation Ste´phanie Massola,n, Nicola Molinaroa,b, Manuel Carreirasa,b a

BCBL – Basque Center on Cognition, Brain and Language, Donostia, Spain IKERBASQUE – Basque Foundation for Science, Bilbao, Spain

b

art i cle i nfo

ab st rac t

Article history:

Two experiments investigated the lexical inhibitory effect of orthographic neighbors

Accepted 31 January 2015

relative to identity priming effects in an unmasked priming paradigm combined with a

Available online 7 February 2015

lexical decision task on word targets. Targets were preceded either by the same word, by a

Keywords:

lower frequency orthographic word neighbor, by an orthographic pseudoword neighbor or

Event-related potentials

by an unrelated prime. Experiment 1 showed a standard facilitatory effect from identity

Lexical access

primes, whereas inhibitory priming effects were observed for both types of neighbor

Orthographic priming

primes. Experiment 2 examined the time-course of these effects by using event-related

Visual word recognition

potential recordings. A generalized relatedness effect was found in the 200–400 ms time-

Lexical competition

window, with smaller negativities generated by related primes than unrelated primes regardless of prime type. In contrast, at 400 ms, while identity primes were associated with smaller negativities than unrelated primes, word neighbor primes were associated with greater negativities than unrelated primes. Additionally, pseudoword neighbor primes produce null effects as compared to unrelated primes. These results are discussed in terms of competition between activated lexical representations and revealed that such a mechanism is modulated by the lexical status of the prime. & 2015 Elsevier B.V. All rights reserved.

1.

Introduction

Successful reading requires translating visual symbols into meaning. Research in cognitive neuroscience has indicated that individual written words can be identified in less than half a second. The speed and apparent ease of this process has provided a continuing challenge for researchers seeking to understand the mechanisms involved in visual word recognition. One point on which there is presently a consensus is that

individual written words are identified via their constituent letters (see Grainger, 2008, for a summary of the arguments). One consequence of this point is that a given word does not activate only the lexical representation of that particular word but also that of orthographically similar words, so-called orthographic neighbors (e.g., lift, list and pint are neighbors of lint; Coltheart et al., 1977). The effects of orthographic similarity have been the focus of several investigations in the past several decades. One central open question concerns

n Correspondence to: Basque Center on Cognition, Brain and Language Paseo Mikeletegi 69, 2nd Floor, 20009 Donostia, Spain. Fax: þ34 943 309 052. E-mail addresses: [email protected] (S. Massol), [email protected] (N. Molinaro), [email protected] (M. Carreiras).

http://dx.doi.org/10.1016/j.brainres.2015.01.051 0006-8993/& 2015 Elsevier B.V. All rights reserved.

36


the structural organization of the mental lexicon and so the present investigation aims to provide further insights into the nature of the connections between lexical representations, as well as between prelexical orthographic and lexical representations. This is accomplished by examining the time-course of visual word recognition using an unmasked priming paradigm combined with event-related potential (ERP) recordings. There is a general consensus nowadays that word recognition involves the selection of the correct lexical representation from among a set of possible candidates (i.e., orthographic neighbors). One prominent model of visual word recognition, the Interactive-Activation model (IA model, McClelland and Rumelhart, 1981), implements competitive interactions between activated lexical representations to ensure a word's recognition while inhibiting its competitors. More precisely, the original IA model is a connectionist model with three hierarchical levels of letter string processing: a feature level (which corresponds to visual segments of letters), a letter level, and a word level. The representational units at each level feed activation to (as applicable) the levels of representation above and below them in the hierarchy, and there is lateral inhibition among representations within each level. According to this model, lexical representations that share letters with a presented word will receive bottom-up support and the orthographic overlap between the presented word and words with similar orthography will modulate the activation of lexical representations. This processing dynamic also implies that words with many orthographic neighbors will produce higher levels of resonance through excitatory connections between the different levels of representation – for instance, those at the letter level and those at word level – than words with few orthographic neighbors (Andrews, 1989, 1992, 1997). To ensure the recognition of a visually presented word, the IA model and its successors (e.g., Davis and Lupker, 2006; Grainger and Jacobs, 1996; Perry et al., 2008) account for competitive interaction in the form of lateral inhibitory connections between activated lexical representations. According to that mechanism, any activated lexical representation spreads inhibition to all other lexical units which allows the suppression of other lexical representations that have received some activation from letter level representations. In a series of experiments using a single word reading paradigms, Grainger and collaborators showed that the size of the set of lexical candidates per se does not affect the selection of the correct representation, but rather that the presence of an activated lexical representation of a higher frequency neighbor of the presented word dominates the word recognition process (Grainger, 1990; Grainger et al., 1989; Grainger and Segui, 1990; Jacobs and Grainger, 1992). Thus, when a given word is presented in isolation, the response times in a lexical decision task are longer when the word possesses a higher frequency orthographic neighbor (see also Carreiras et al., 1997; Grainger, 1990; Perea and Pollatsek, 1998). According to the IA approach, each representation has a threshold of recognition which depends on the lexical frequency of the stimulus. At the resting state, this threshold is lower for high-frequency words and higher for low-frequency words. Consistent with this characteristic of the IA framework, several studies have reported faster

reaction times and better accuracy for high-frequency words than for low-frequency words (e.g., Frost et al., 1987; Grainger, 1990; Humphreys et al., 1988; Perea and Carreiras, 1998). Once past its threshold, a lexical representation sends inhibitory input to all other lexical representations as a function of its activation level and so representations of high-frequency words produce greater inhibition than representations of low-frequency words. To date, the majority of evidence regarding orthographic neighborhood effects comes from studies using the masked priming paradigm, in which a briefly presented prime (typically on the order of 50 ms in duration) is followed by a target item (Forster and Davis, 1984). Numerous studies have investigated priming effects from identity and neighbor primes. Overall, these studies show that word prime that is orthographically related to a lower frequency target tends to interfere with target processing (Carreiras et al., 1997; Davis and Lupker, 2006; De Moor and Brysbaert, 2000; Nakayama et al., 2008; Perea and Rosa, 2000a; Segui and Grainger, 1990), whereas the presentation of the same word or an orthographically related pseudoword as prime stimulus facilitates target processing relative to unrelated controls (Bodner and Masson, 1997; Forster, 1987; Forster and Davis, 1984; Forster et al., 1987, 2003; Forster and Veres, 1998; Grainger and Jacobs, 1993; Perea and Lupker, 2003, 2004; Perea and Rosa, 2000a; Sereno, 1991). According to the IA framework, the activation generated during prime processing is maintained upon presentation of the target stimulus and so accumulated activation in the network will produce initial facilitation when prime and target are orthographically similar, facilitation being maximal when the target is the same word as the prime stimulus. However, when the prime is a neighbor of a lower frequency target, the prime supports the activation of a lexical representation other than the representation of the target stimulus and so within-level inhibition on the word target will develop. The amount of inhibition will be maximal when primes are of higher frequency than the word targets. Consequently, the word neighbor prime is capable of substantial inhibition of its neighbors and this lengthens the recognition of the word target. Recent research has combined the masked priming paradigm with event-related potentials (ERPs) to provide complementary data, particularly with respect to the relative timing of effects found in behavioral experiments. ERPs have been extensively used for studying visual word recognition, because this technique enables the temporal ordering (with millisecond-level resolution) of the neural processes involved during reading. Therefore this technique allows the dissociation of hierarchically different sources of information (such as orthographic, semantics) through a continuous measure of neuronal activity over time. Combined with the masked priming paradigm, this methodology has been shown to be sensitive to different levels of processing occurring during visual word identification, such as orthographic processing (Carreiras et al., 2009a, 2009b; Grainger et al., 2006; Holcomb and Grainger, 2006), lexical processing (Carreiras et al., 2009a, 2009b; Holcomb and Grainger, 2006; Massol, 2012; Massol et al., 2011), and semantic processing (Holcomb and Grainger, 2006; Midgley et al., 2009). Most relevant to the present investigation is the study conducted by Massol et al. (2010), in which they investigated


effects of briefly presented primes that were orthographic neighbors of low-frequency target words. The effects of word neighbor primes were compared with the effects of repetition primes and with the effects of pseudoword neighbor primes. In an early time-window, from 175 ms to 300 ms post-target onset, all types of orthographically related primes showed a similar priming effect, with smaller negativities relative to targets following an unrelated prime. In contrast, between 300 ms and 550 ms, repetition primes and pseudoword neighbor primes generated more positive-going waveforms than unrelated primes, whereas there was no difference between targets following a word neighbor prime and targets following an unrelated word prime. This null effect suggests that facilitatory effects of orthographic overlap, as highlighted by the reduced negativity found with pseudoword neighbor primes, were offset by an inhibitory influence of the prime, because the lexical representations of the prime and target are competing during word target recognition. These results have been extended and recent studies reported an even greater N400 for low-frequency targets following a word neighbor prime than targets following an unrelated word prime (Gobin et al., 2012; Massol, 2012), thus supporting the hypothesis of lateral inhibition from activated lexical representations. One limitation of the masked priming technique is that it usually involves a very short interval between prime and target stimuli. Consequently, this technique is most useful for investigating the initial stages of visual word recognition and thus early automatic processes (Forster and Davis, 1984; Grainger, 2008; Kinoshita and Lupker, 2003). In contrast, when the prime is totally unmasked (and presented for at least 200 ms), there is conscious processing of the prime before the target presentation. Several studies have highlighted qualitative differences between masked and unmasked priming paradigms (Dehaene et al., 2001; Forster et al., 2003; Grainger, 2008). For instance, regarding identity priming effects, stronger effects are obtained for low-frequency words than for highfrequency words when the prime is unmasked, whereas such differences are usually not observed under masked conditions (Forster and Davis, 1984; Humphreys et al., 1988; Versace and Nevers, 2003; but see Grainger et al., 2012). Moreover, Perea and Rosa (2000b) reported that the magnitude of the unmasked identity priming is larger relative to the masked identity priming. Thus, it has been suggested that effects from unmasked priming can be also mediated by an episodic memory trace of the prime stimulus (Balota et al., 2008; Bodner and Masson, 2003; Forster and Davis, 1984; Versace, 1998; Versace and Nevers, 2003) or due to changes in the decision processes (Gomez et al., 2013; Kinoshita et al., 2011). Although there are different interpretations regarding the identity priming effects in unmasked priming paradigm, the present study was designed to contrast identity priming effects with the effects from orthographically related primes and to provide further electrophysiological evidence of these neighborhood effects. Regarding orthographic neighborhood effects, studies typically report inhibitory effects when targets are preceded by an unmasked related prime (Burt, 2009; Colombo, 1986; Gobin and Mathey, 2010; Robert and Mathey, 2007; Segui and

37

Grainger, 1990). The first evidence for this claim comes from Colombo's (1986) study, which showed that inhibitory effects from orthographic neighbor primes are found when the prime is of lower frequency than the target (see also De Moor and Verguts, 2006). This pattern of findings was taken as evidence for a side-effect of lexical competition between activated representations during prime word identification. Segui and Grainger (1990) extended these results and showed that these effects are not exclusively due to the frequency of the word targets, but are instead due to the relative frequency between prime and target words. Low-frequency neighbor primes, with a 300 ms duration, slowed down performance to word targets (e.g., axle-ABLE) relative to unrelated primes, as reflected by an increased in lexical decision latencies and error rates, whereas no significant priming effect was observed when the primes were of higher frequency than the target (e.g., able-AXLE). These results, which are in the opposite direction to those from masked priming studies, suggest that because selection of the lexical representation of the prime has occurred before the target presentation, all high-frequency neighbors of the prime have been removed from competition. This prior rejection of the other competitors has direct consequences for the target identification, by increasing the competition between the lexical representations of the prime and the target when the target appears. More recently, a few studies have directly compared the orthographic neighbor effects of a masked prime with those of an unmasked prime (Gobin et al., 2012; Gobin and Mathey, 2010; see also Perea and Rosa, 2000a, 2000b). For instance, Gobin and Mathey (2010) reported an inhibitory priming effect that was enhanced by longer prime duration (i.e., 66 ms vs. 166 ms for prime durations). The authors argued that the increased duration of the prime reinforces the competitiveness between lexical candidates during target identification. Gobin and colleagues (2012) extended these results by providing ERP evidence of whether the time-course of word neighbor priming effect can be modulated by prime duration. Whereas manipulating the prime duration leads to qualitatively similar influence of the word neighbor prime, greater priming effects were found with 166 ms prime duration. These results are consistent with previous behavioral findings on neighbor priming effects (Burt, 2009; Gobin and Mathey, 2010; Perea and Rosa, 2000a, 2000b) as well as with previous ERP findings on repetition priming effects (Holcomb and Grainger, 2007; Holcomb et al., 2005). Then, these recent studies highlight that the inhibitory effect of unmasked neighbor primes is thought to occur as a side effect of the resolution of competition required for prime identification, due to a change of activation of lexico-semantic representations prior to the target's presentation (see also Frost et al., 1997; Grainger, 1990; Grainger and Ferrand, 1996; Lukatela and Turvey, 1994; Neely, 1991). Complicating matters further, some researchers have noted that the inhibitory effect from orthographic neighbor primes is not restricted to pairs of words. Versace (1998) showed that pseudoword neighbor primes produced an inhibitory effect on word target identification (see also Burt, 2009; Colombo, 1986, for similar results). This was interpreted as suggesting that unmasked priming effects are the result of the sum of a residual activation component and retrieval of

38


an episodic memory trace (see also Durgunoglu and Neely, 1987; Forster et al., 1990; Forster and Davis, 1984; Humphreys et al., 1988; Versace and Nevers, 2003; Whitlow, 1990; Whitlow and Cebollero, 1989). Consequently, the inhibitory effects from orthographic neighbor primes – both words and pseudowords – were theorized to reflect a long-term process. According to Versace (1998), prime processing leads to the construction of an orthographic (word and pseudoword) representation, which modifies the content of long-term memory and does not simply modify the activation of lexical representations. Recently, Burt (2009) investigated whether this mechanism underlies orthographic neighbor effects in both masked and unmasked priming paradigms. She found that word neighbor primes consistently produced interference during target identification, whereas pseudoword neighbor primes were associated with inhibitory effects only when targets were high-frequency words. She therefore showed that there was no evidence for selective inhibition of targets by lower frequency primes, and so argued that masked and unmasked orthographic priming are driven by a common mechanism. Thus, to date there is no consensus about whether the effects from masked and unmasked primes are influencing common or different levels of representation and/or mechanisms. For this reason, the present study focused on whether unmasked orthographically related primes generate competition during word target processing. Additionally, it aimed to provide further evidence regarding the source of the effects of orthographic overlap between prime and target by combining an unmasked priming paradigm with ERP recordings. More specifically, it aimed to determine whether the previously discussed inhibitory effects are a consequence of activated lexical representations from orthographic neighbors of the target, or whether these effects can be modulated by the lexical status of the unmasked prime. According to some researchers (Burt, 2009; Versace, 1998; Versace and Nevers, 2003), prime identification consists of inhibiting all other lexical candidates and in particular, those of higher frequency. To this extent, when a prime is a pseudoword, all lexical representations of orthographic neighbors have to be inhibited during prime identification and the degree of inhibition is a function of both the similarity to the prime and the neighbor frequency. Thus, when the word target is presented, its identification should be lengthened because of the previous inhibition of lexical representations during prime processing, just as is the case for lower frequency neighbor primes of the word target. Another possibility is that orthographic neighbor effects obtained in unmasked priming paradigm are modulated by the lexical status of the prime. Activation mechanisms at prelexical and lexical levels are thought to be automatic and short-lived and so such activation is assumed to rapidly decrease if no reactivation takes place (Versace and Nevers, 2003). The amount of activation at the lexical level produced by a pseudoword stimulus should be still weaker than if the stimulus is a word. Consequently, this weak activation leads to no or little competition among the lexical representations, and so these effects can be considered as short-lived. In contrast, activation of lexico-semantic information is thought to have a long-term effect and so the more the lexical

representations are activated by the stimulus, the more time it will take to these representations to fully return to their resting state. Thus, identification of a word prime leads to a higher level of activation of its lexical representation as compared to other representations, and so inhibition from this representation can affect word target identification. In light of the prior studies cited above, the explanation of the locus of effects from an orthographic neighbor prime remains a topic of controversy. The present investigation focuses on the locus of the neighbor priming effects in unmasked priming paradigm. To do so, effects of lower frequency orthographic neighbor primes are compared to effects of identity primes and effects of pseudoword neighbor primes. The identity priming condition provides a focus on bottom-up influences from an orthographically related prime on target processing and so helps to tease apart priming effects from orthographic overlap from effects of lower frequency primes that should occur at a lexical level of processing. As an improvement over prior research, the lexical status of the neighbor prime was also manipulated, while keeping orthographic overlap constant to provide a further means of examining lexical influences of neighbor primes. Experiment 1 was aimed at providing insight into this issue by using a lexical decision task on target stimuli. Experiment 2 aimed to investigate the time-course of such effects using electrophysiological recordings (ERPs). In sum, the main goal of the present investigation is to provide evidence on the temporal dynamics of orthographic and lexical influences from unmasked orthographic neighbor priming during word target processing.

2.

Experiment 1: behavioral experiment

In Experiment 1, effects of word neighbor priming will be compared with both identity priming effects and effects of pseudoword neighbor priming. Based on the IA account of inhibitory priming effects from orthographic neighbors as reflecting competitive interactions between activated representations at the lexical level, and based on previous evidence, we expect to see a divergence between effects from neighbor primes and effects from identity primes. According to previous studies, targets preceded by a neighbor prime should be associated with longer latencies than targets preceded by an unrelated prime (see Burt, 2009, among others), whereas identity priming should produce a facilitatory effect (see Perea and Rosa, 2000b). Participants were presented with visual primes that were displayed for 200 ms (see Perea and Rosa, 2000b, for a similar prime duration) and directly followed by a target. They were instructed to decide as rapidly and as accurately as possible whether or not the target was a Spanish word. The critical word targets were preceded either by the same word (e.g., boca-BOCA [mouthMOUTH]), or by a lower frequency orthographic neighbor word (e.g., beca-BOCA [grant-MOUTH]), or by an orthographic pseudoword neighbor (e.g., bica-BOCA [MOUTH]), or by an unrelated prime. Each related prime was paired with an unrelated prime which was either a low-frequency word (e.g., arte-BOCA [skill-MOUTH]), a high-frequency words

39


Table 1 – Mean reaction times (in ms) and percentages of errors (in italic) to word targets preceded by different types of prime in Experiment 1 (standard deviations are in parenthesis). Effect is the outcome of the subtraction of the values in the unrelated condition from those in the related condition. Type of prime

Identity Word neighbor Pseudoword neighbor

Priming

RT Error RT Error RT Error

(e.g., arte-BOCA [skill-MOUTH], or a pseudoword (e.g.,arceBOCA [maple-MOUTH]).

2.1.

Results

Statistical analyses were performed only over the trials which involved words in target position. Incorrect responses and reaction times below 250 ms and above 1500 ms (.21% of the data) were excluded from the latency analysis. Mean latencies for correct responses and error rates are presented in Table 1. ANOVAs on these latencies and on error rates were performed over participants and items with Type-of-Prime (Identity, Word neighbor, Pseudoword neighbor) and Relatedness (Related, Unrelated) as main factors. Relevant main effects were further evaluated by ANOVAs with the same design, comparing the conditions in a pairwise manner.

2.1.1.

Reaction times

There was a significant main effect of Type-of-Prime, F1(2,76)¼ 48.93, MSE¼508.56, po.001; F2(2,454)¼23.52, MSE¼ 8018.02, po.001. There was no main effect of Relatedness (F1(1,38)¼.20, MSE¼ 531.72, p¼ .652; F2(1,227)¼ .25, MSE¼4296.23, p¼.875). Importantly there was a significant Type-of-Prime Relatedness interaction, F1(2,76)¼38.61, MSE¼655.31, po.001; F2(2,454)¼ 30.51, MSE¼5509.13, po.001. Follow-up analyses revealed significantly longer reaction times when word targets were following a word neighbor prime than when word targets were following an unrelated word, F1(1,38)¼ 15.12, MSE¼ 921.93, po.001; F2(1,227)¼ 20.76, MSE¼ 4983.24, po.001, whereas shorter reaction times were observed when word targets were identical to the prior prime than when word targets were preceded by an unrelated word, F1(1,38)¼77.68, MSE¼ 442.33, po.001; F2(1,227)¼ 43.40, MSE¼ 4821.28; po.001. Furthermore, the effect of relatedness was also significant for pseudoword neighbors (F1(1,38)¼ 5.04, MSE¼ 478.08, p¼ .031; F2(1,227)¼4.28, MSE¼ 5509.99, p¼ .040). Longer reaction times were observed for targets following a pseudoword neighbor prime as compared to targets following an unrelated pseudoword prime. Finally, the three unrelated conditions did not differ from each other (p4.1).

2.1.2.

Error rates

An ANOVA on the error data for word targets revealed a significant main effect of Type-of-Prime, F1(2,76)¼8.30, MSE¼ 26.09, p¼ .001; F2(2,454)¼13.63, MSE¼ 104.92, po.001 and a significant main effect of Relatedness (F1(1,38)¼6.70,

Effect

Related

Unrelated

591 5.19 655 9.31 650 10.66

633 6.61 628 6.54 639 7.76

(124) (5.64) (129) (6.15) (120) (9.52)

(123) (7.24) (116) (7.41) (119) (7.11)

42 1.42 27 2.77 11 2.90

MSE¼ 17.53, p¼ .014; F2(1,227)¼ 7.34, MSE¼105.24, p¼.007). The Type-of-Prime Relatedness interaction was also significant, F1 (2,76)¼6.99, MSE¼ 16.80, p¼ .002; F2(2,454)¼ 6.26, MSE¼113.61, p¼ .002. Follow-up analyses revealed significantly more errors when targets followed a word neighbor prime than when word targets followed an unrelated word, F1(1,38)¼ 5.89, MSE¼ 25.32, p¼ .020; F2(1,227)¼ 7.73, MSE¼132.15, p¼ .006, and when targets followed a pseudoword neighbor prime relative to targets following an unrelated pseudoword prime, F1(1,38)¼ 10.88, MSE¼ 15.08, p¼ .002; F2(1,227)¼7.99, MSE¼ 119.58, p¼.005. However the effect of relatedness was marginal for identity primes in the analysis by subject, F1(1,38)¼3.65, MSE¼10.72, p¼.064; F2 (1,227)¼ 2.69, MSE¼80.73, p¼ .102. Finally the three unrelated condition did not significantly differ from each other (p4.1).

2.2.

Discussion

The results of the Experiment 1 are clear-cut. First of all, the results replicated the standard findings of inhibitory effects from low-frequency orthographic neighbor primes in the lexical decision task combined with unmasked priming paradigm (e.g., Burt, 2009; Colombo, 1986; De Moor and Verguts, 2006; Segui and Grainger, 1990; Versace, 1998). When the word neighbor is of lower frequency than the word target, target identification is slowed down relative to an unrelated condition. Second, targets following a pseudoword neighbor prime were also associated with significantly longer latencies and more errors than following an unrelated pseudoword prime. Finally, a facilitatory effect was found, only on the reaction time data, when target words were an exact repetition of the prime stimulus as compared with word targets preceded by an unrelated prime. The IA framework postulates that lexical representations activated by a stimulus do not immediately deactivate but rather their activation decays over time to get back to their resting state level. Therefore, when the target is a repetition of the word prime, less processing for target identification is required due to the residual activation, resulting in shorter lexical decision latencies as well as more accurate responses (see Bowers, 2000, 2003; Forbach et al., 1974; Forster and Davis, 1984; Monsell, 1985; Rajaram and Neely, 1992, among others). In other words, the identity prime helps the lexical process to select a single lexical representation, due to the continuous interaction between different levels of representations during the target processing (McClelland and

40


Table 2 – Mean reaction times (in ms) and percentages of errors (in italic) to word targets preceded by different types of prime in Experiment 2 (standard deviations are in parenthesis). Effect is the outcome of the subtraction of the values in the unrelated condition from those in the related condition. Type of prime

Identity Word neighbor Pseudoword neighbor

Priming

RT Error RT Error RT Error

Rumelhart, 1981). Because the representations are shared by the prime and target stimuli, identity priming relies mainly on pre-activation of orthographic and lexical representations (but see Balota et al., 2008; Bodner and Masson, 2003; Forster and Davis, 1984; Gomez et al., 2013; Kinoshita et al., 2011; Versace, 1998; Versace and Nevers, 2003 for different interpretations for identity priming with unmasked primes). The fact that an inhibitory effect from neighbor priming has been observed strongly suggests that the lexical selection for identifying the prime stimulus has direct consequences on word target identification. Identification of the prime stimulus involves an elimination of inappropriate candidates until it is recognized. As suggested by Colombo (1986) and by Segui and Grainger (1990), this selection process may require the active inhibition of similar activated lexical representations, especially lexical representations of higher frequency words than the prime stimulus. Consequently, during target processing, this prior rejection increases the competition between lexical representations activated by both prime and target stimuli, resulting in an inhibitory effect from neighbor priming. In addition, the present results also revealed that participants took more time and made more errors to target words following a pseudoword neighbor prime relative to target words following an unrelated pseudoword. Even though a pseudoword has no lexical representation, the pseudoword prime activates lexical representations, as a word prime does. However, the lexical selection cannot be completed because there is no lexical entry that exactly matches the prime. Consequently those representations will stay slightly activated over time, until they return to their resting state. Thus, this activation of competitors will still lead to competition among the lexical representations during target recognition. Summing up, the inhibitory effect from neighbor priming is the consequence of a within-level competition between activated lexical representations, whereas the identity priming effect is driven by the activation of shared orthographical and lexical representations between prime and target stimuli. Experiment 2 aims to provide electrophysiological evidence for these mechanisms and their time-courses. The broad literature on ERP effects on visual word recognition has highlighted that effects arising around 200 ms are likely to be modulated by orthographic factors, such as bigram frequency (Laszlo and Federmeier, 2014), orthographic regularities (Bentin et al., 1999; Coch and Mitra, 2010), and orthographic overlap between masked prime and target stimuli (Carreiras et al., 2009a, 2009b; Holcomb and Grainger, 2006; Massol et al., 2010, 2012). In the subsequent interval, starting around

Effect

Related

Unrelated

578 4.15 641 11.40 635 12.28

623 6.62 620 6.53 626 8.45

(68) (4.69) (77) (8.98) (76) (8.44)

(73) (5.99) (71) (8.42) (70) (8.77)

45 2.47 21 4.87 9 3.83

400 ms, it has been reported that ERP waveforms can be modulated by repetition priming (Chauncey et al., 2008; Holcomb and Grainger, 2007), by semantic integration (Molinaro and Carreiras, 2010), among other lexico-semantic factors. All this evidence suggests that starting around 400 ms there is activation at the lexical and semantic level as a tentative to map one representation as unique interpretation of the presented stimulus, as well as to map activated lexical representations onto meaning due to processing dynamic as described in the IA model (McClelland and Rumelhart, 1981). As for the identity priming effect, we expect to find early and late ERP effects, reflecting orthographic prelexical processing and lexical processing respectively. Regarding the neighbor priming effects, we expect to find some differences between word and pseudoword primes arising at the lexical level. If primes trigger first facilitation followed by inhibition of word candidates only when the prime is a word, we predict similar effects for word and pseudoword primes in early components followed by a lexical effect arising around 400 ms from word neighbor primes. In other words, similar effects from identity priming and neighbor priming should be observed before 400 ms post-target onset due to the orthographic overlap between prime and target stimuli. We also expect to see a divergence in these effects arising around 400 ms, thus reflecting competition between activated lexical representations in order to map one representation as unique interpretation of the presented stimulus.

3.

Experiment 2: ERP experiment

3.1.

Results

3.1.1.

Behavioral data

Statistical analyses were performed only over the trials which involved words in target position. Incorrect responses and reaction times below 250 ms and above 1500 ms were excluded from the latency analysis (2.40% of the data). Mean latencies for correct responses and error rates are presented in Table 2. ANOVAs on these latencies and on error rates were performed over participants and items with Type-ofPrime (Identity, Word neighbor, Pseudoword neighbor) and Relatedness (Related, Unrelated) as main factors. Relevant main effects were further evaluated by ANOVAs with the same design, comparing the conditions in a pairwise manner.


41

Fig. 1 – ERPs time locked to target onset in two conditions (black line: word neighbor, red line: unrelated) over nine electrode sites in Experiment 2.

3.1.1.1. Reaction times. There was a significant main effect of Type-of-Prime, F1(2,64)¼35.21, MSE¼550.67, po.001; F2(2,454)¼ 40.14, MSE¼ 3992.15, po.001. There was a marginal effect of Relatedness, F1(1,32)¼4.10, MSE¼ 358.20, p¼ .051; F2(1,227)¼ 3.12, MSE¼ 3167.13, p¼.079. Critically, the Type-of-Prime Relatedness interaction was significant, F1(2,64)¼ 39.41, MSE¼ 521.63, po.001; F2(2,454)¼ 36.78, MSE¼ 4296.72, po.001. Followup analyses revealed significantly longer reaction times when word targets followed a word neighbor prime than when word targets followed an unrelated word, F1(1,32)¼10.94, MSE¼ 672.30, p¼ .002; F2(1,227)¼11.30, MSE¼ 4143.12, p¼ .001, whereas shorter reaction times were observed when word targets were identical to the prior prime than when word targets were preceded by an unrelated word, F1(1,32)¼ 99.98, MSE¼341.68, po.001; F2(1,227)¼ 66.55, MSE¼ 3960.68, po.001. Regarding the pseudoword neighbor primes a significant effect was found only in the analyses by items, F1(1,32)¼ 2.75, MSE¼387.49, p4.1; F2(1,227)¼4.25, MSE¼3656.77, p¼.040, suggesting that word targets preceded by a pseudoword neighbor primes were associated with longer latencies than word targets preceded by an unrelated pseudoword. Finally, the three unrelated condition did not differ from each other (p4.1). 3.1.1.2. Error rates. An ANOVA on the error data for word targets revealed a significant main effect of Type-of-Prime, F1(2,64)¼ 10.97, MSE¼39.74, po.001; F2(2,454)¼24.31, MSE¼ 130.10, po.001 and a significant main effect of Relatedness (F1 (1,32)¼12.72, MSE¼ 16.72, p¼.001; F2(1,227)¼ 9.96, MSE¼ 147.20, p¼ .002). The Type-of-Prime Relatedness interaction was also significant, F1(2,64)¼13.14, MSE¼ 19.78, po.001; F2(2,454)¼11.96,

MSE¼ 140.04, po.001. Follow-up analyses revealed significantly more errors when word targets followed a word neighbor prime than when word targets followed an unrelated word, F1(1,32)¼ 21.44, MSE¼18.20, po.001; F2(1,227)¼ 15.28, MSE¼166.31, po.001, whereas there were fewer errors when word targets were identical to the prior prime than when word targets followed an unrelated word, F1(1,32)¼ 5.68, MSE¼ 17.73, p¼ .023; F2 (1,227)¼6.72, MSE¼ 91.58, p¼.010. Furthermore, participants also made more errors when word targets followed a related pseudoword neighbor prime than when word targets followed an unrelated pseudoword (F1(1,32)¼11.87, MSE¼ 20.36, p¼ .002; F2 (1,227)¼9.79, MSE¼ 169.41, p¼ .002). Finally, the three unrelated condition did not significantly differ from each other (p4.1).

3.1.2. Electrophysiological measures 3.1.2.1. 200–400 ms post-target epoch. Plotted in Figs. 1–3 are the ERPs contrasting the conditions with related and unrelated primes for word neighbor priming (Fig. 1), for identity priming (Fig. 2) and for pseudoword neighbor priming (Fig. 3). As can be seen in these Figures, between 200 and 400 ms, target words following unrelated primes were associated with a larger negativity than target words following the same prime word (i.e., repetition) or a prime that differed by a single letter (i.e., orthographic neighbor) (see also Fig. 4). This observation was confirmed by a main effect of Relatedness in the midline analysis, F(1, 32)¼5.17, MSE¼5.70, p¼ .030, and by the presence of an effect of Relatedness that interacted with the Anterior–Posterior factor in the analysis including the four lateralized groups, F(1, 32)¼ 7.82, MSE¼ 1.03, p ¼.009. The Relatedness Anterior–Posterior interaction reflects the fact

42


Fig. 2 – ERPs time locked to target onset in two conditions (black line: identity, red line: unrelated) over nine electrode sites in Experiment 2.

Fig. 3 – ERPs time locked to target onset in two conditions (black line: pseudoword neighbor, red line: unrelated) over nine electrode sites in Experiment 2.


43

the left posterior electrode sites than over the right posterior sites (left sites: F(2,64)¼22.26, MSE¼ 2.13, po.001; right sites: F (2,64)¼ 13.59, MSE¼2.19, po.001). These interactions were driven by the fact that target words following a word neighbor prime were associated with a larger negativity compared with target words following an unrelated prime (F(1,32)¼ 6.15, MSE¼ 2.58, p¼ .019), whereas target words following identity primes were associated with a reduced negativity compared with target words following an unrelated prime (F(1,32)¼ 35.56, MSE¼ 2.42, po.001). The effect of pseudoword neighbor primes was not significant at any of the electrode configurations (all p4.1).

Fig. 4 – Voltage maps centered on the two epochs used in the statistical analyses. The maps represent voltage differences at each electrode site calculating by subtracting the voltage values in the related prime condition from the voltage values in the corresponding unrelated prime condition in Experiment 2.

that the effect of Relatedness was significant at the anterior electrode sites but not at the posterior ones (anterior sites: F (1, 32)¼5.39, MSE¼3.31, p ¼.027; posterior sites: F(1, 32)¼ .02, MSE¼2.02, p¼ .888). Although Fig. 4 shows that the effects of identity primes appear to be more widespread than the effects of orthographic neighbor primes in this time window, the critical two-way interaction between Relatedness Typeof-Prime was not significant and did not interact with any of the electrode configurations (all p4.1).

3.1.2.3. Time-course analyses. Here we examine the timecourse of priming effects of identity, word neighbor and pseusodoword neighbor primes using 50 ms time bins in order to provide a more fine-grained analysis of the evolution of these effects over time (Table 4). We present the timecourse of effects of the three type of priming separately, given that there was a significant interaction in the 400–600 ms time-window. The table represents the significant effects (corrected p-values) obtained with ANOVAs in each of the nine successive time-windows. The results of the time-course analyses confirm and extent the results obtained in the epoch-based analyses presented above. Clear effects of identity priming were first evident in the 300–350 ms time-window (Identity Anterior–Posterior: F (1, 32)¼13.65, MSE¼ 1.90, p¼ .001; Anterior sites: F(1, 32)¼6.36, MSE¼5.53, p¼ .017; Posterior sites: F(1, 32)¼ .43, MSE¼3.78, p¼ .516), and continued to have an influence up to 600 ms post-target onset. Effects of word neighbor primes were seen between 400 and 600 ms post-target onset, whereas effects of pseudoword primes were only significant in the 400–450 ms time-window, F(1, 32)¼ 6.34, MSE¼ 8.13, p¼ .017. 3.2.

Discussion

3.1.2.2. 400–600 ms target epoch. In the 400–600 ms timewindow, targets following word neighbor primes produced a more negative-going wave than targets following unrelated primes, whereas targets following identity primes produced a less negative-going wave in this epoch than word targets following unrelated primes (see Figs. 1 and 2). However, as can be seen in Fig. 3, the difference between targets following unrelated and pseudoword neighbor primes were quite small. Altogether, this reflects the widespread nature of repetition and word neighbor priming effects in this time-window, and a notable absence of priming effects from a pseudoword neighbor. These observations were confirmed by the presence of a significant Relatedness Type-of-Prime interaction in the lateralized groups analysis, F(2,64)¼ 16.92, MSE¼7.29, po.001, as well as in the midline analysis, F(2,64)¼ 18.03, MSE¼ 9.98, po.001 and also a significant Relatedness Type-of-Prime Anterior–Posterior Hemisphere interaction, F(2,64)¼ 4.09, MSE¼.07, p¼.026. Follow-up analysis revealed that the interaction between Relatedness, Type-of-Prime and Hemisphere was significant only over posterior sites (anterior sites: F(2,64)¼ 1.21, MSE¼ .28, p4.1; posterior sites: F(2,64)¼ 5.77, MSE¼.18, p¼ .006). Analyses also revealed that the Relatedness Type-of-Prime interaction was significantly larger over

At the behavioral level, this pattern of results almost perfectly1 replicated the results observed in Experiment 1, showing facilitatory effect from identity primes and an inhibitory priming effect from both types of neighbor prime. Altogether, these results are consistent with the previous data reported in the literature (Burt, 2009; Segui and Grainger, 1990; Versace and Nevers, 2003; among others). At an electrophysiological level, a generalized Relatedness effect was found between 200 ms and 400 ms post-target onset, mainly distributed over the most anterior electrode sites. In the subsequent epoch, between 400 ms and 600 ms, identity primes were associated with smaller negativity relative to unrelated primes, whereas targets preceded by a word neighbor prime were associated with larger negativities relative to targets preceded by an unrelated prime. In this time-window, pseudoword neighbors did not differ from the unrelated 1 Although the priming effect of word neighbor prime was significant only in the analyses by items in Experiment 2 but significant in both analyses by subjects and by items in Experiment 1, the numerical trend is consistent across studies. to the same is true for the priming effects from pseudoword neighbor primes.

44


condition. The present results are in line with the results observed in Experiment 1 and provide further evidence about the temporal dynamics of orthographic and lexical influence from unmasked orthographic neighbor primes during target processing, by showing clear effects at two different timewindows. Altogether, these effects strongly suggest that the lexical status of the prime starts to influence word target processing at around 400 ms. The first of the observed effects takes place between 200 and 400 ms post-target onset, with larger negativity for unrelated targets as compared to targets preceded by a related prime regardless of the prime type (i.e., identity, word neighbor, pseudoword neighbor). However, it should be noted that this effect seems to be mainly driven by the identity condition (see Fig. 4), even if the statistical analyses did not reveal any significant interaction between the two factors. As described in the Introduction, unmasked primes can be consciously perceived by the participant and fully recognized before the presentation of the target stimulus. Selected processes at the lexical level have been completed during word prime processing and so lead to the activation of some aspects of lexico-semantic knowledge prior to the target's presentation. Consequently, right before target onset, there are transient variations in the activation level of lexical representations (Frost et al., 1997; Grainger and Ferrand, 1996; Lukatela and Turvey, 1994; Neely, 1991) and these transient variations can affect target recognition. Moreover, the presentation of the orthographically related target leads to a reactivation (at least partly) of some pre-lexical and lexical representations (Holcomb and Grainger, 2007; Holcomb et al., 2005; Grainger and Holcomb, 2009). According to the IA model (McClelland and Rumelhart, 1981), activated lexical representations feed back activation to representations at the letter level, and so the orthographic overlap between prime and target is associated with a reactivation of some of the pre-lexical representations. Thus, the widespread effect from identity priming can be simply due to the full match in terms of orthographic features between the activated lexical representation of the prime and target stimulus. In the second time-window, between 400 ms and 600 ms post-target onset, ERPs revealed an identity priming effect, with larger negativity for unrelated targets than for related targets, an opposite pattern for word neighbor priming and no priming effect from a pseudoword neighbor. These results are consistent with previous findings (Holcomb and Grainger, 2007; Holcomb et al., 2005; see also Gobin et al., 2012,2). These effects strongly support the hypothesis of competition between all activated lexical representations of orthographically similar words of both prime and target stimuli. However, the results clearly revealed that the lexical status of the prime does modulate such mechanism, as discussed in Section 4.

2 Even though Gobin et al. (2012) did not report statistical analyses in the 400–600 ms time-window, the figures revealed differences between word neighbor primes and unrelated primes. These previous data showed a larger negativity associated with targets that were preceded by a word neighbor prime relative to targets preceded by an unrelated prime.

4.

General discussion

The present investigation aimed to explore the orthographic and lexical influences provided by an unmasked orthographic neighbor prime on word target processing relative to repetition priming effects. Behavioral results from both experiments revealed a facilitatory effect from identity priming, whereas orthographic neighbor primes produced inhibitory effects on target processing regardless of their lexical status. This pattern of results replicated previous findings by showing that identifying an orthographically related prime produces inhibition of word targets (Burt, 2009; Colombo, 1986; De Moor and Verguts, 2006; Gobin and Mathey, 2010; Segui and Grainger, 1990). Experiment 2 extended these results by providing a finer temporal analysis of these effects by combining the lexical decision task with ERPs. On the basis of the results found in Experiment 1 and prior work using ERPs, we predicted specific patterns of priming for identity, word neighbors and pseudoword neighbors in ERP epochs reflecting the lexico-semantic processing, with an interaction between Relatedness and Type-of-Prime starting approximately at 400 ms post-target onset. The results of Experiment 2 are in line with this prediction. In the 200–400 ms time-window, smaller amplitudes were observed when targets followed a prime that was orthographically related (identity or orthographic neighbor) regardless of the lexical status of the prime as compared to targets that did not have any letters in common with the prime. Therefore, effects occurring in this time-window seem to reflect target processing at the interface between orthographic prelexical and lexical representations (Grainger and Holcomb, 2009; Laszlo and Federmeier, 2014). According to these previous studies using different paradigms and approaches (Bentin et al., 1999; Carreiras et al., 2009a, 2009b; Coch and Mitra, 2010; Holcomb and Grainger, 2006; Massol et al., 2010, 2012; McCandliss et al., 2003; Tarkiainen et al., 1999), the effects observed between 200 ms and 400 ms are hypothesized to reflect orthographic processing and/or the mapping of prelexical orthographic representations onto lexical representations. In the present investigation, because prime recognition occurred before target presentation, some previously activated lexical representations did not fully return to their resting states, when the target stimulus appeared. Consequently, these remaining activated lexical representations send excitatory feedback to all compatible letter representations that are activated by the visual stimulus (i.e., here the target stimulus). Target processing is improved when the latter shares some letters with the prime; this compatibility being maximal when the target is the exact repetition of the word prime (Forster and Davis, 1984; Holcomb and Grainger, 2006; Perea and Rosa, 2000a, 2000b). Indeed, ERP effects from identity priming suggest that representations that were activated by the prime stimulus are receiving activation from the target. This produces higher levels of resonance through excitatory connections between the different levels of representation, especially those at the letter level and those at the lexical level. The key result of the present study concerns the significant interaction between the Type-of-Prime and Relatedness


factors seen in the 400–600 ms time-window. Whereas the behavioral data highlighted inhibitory effects from both word and pseudoword neighbor primes and a facilitatory effect of identity priming, the ERPs revealed the effects were different in the three priming conditions in the 400–600 ms timewindow. On the one hand, smaller negativity was produced by identity priming as compared to the unrelated condition (see Holcomb et al., 2005, for similar results). On the other hand, targets following a word neighbor prime were associated with a larger negativity than targets following an unrelated prime, whereas no evidence of pseudoword neighbor priming was found. According to previous results, effects observed in this time-window are hypothesized to reflect the interface between lexical representations and semantic representations (see Grainger and Holcomb, 2009). For instance, it has been demonstrated that an increase of the N400 amplitude reflects the amount of effort involved in forming links between lexical and semantic information (Grainger and Holcomb, 2009; Kutas and Hillyard, 1980). Firstly, words of high-frequency are associated with a smaller negativity than words of low-frequency (Barber et al., 2004; Bentin et al., 1985; Neville et al., 1992; Rugg, 1990; Van Petten and Kutas, 1990). According to the interactive-activation approach, lexical representations of highly frequent words have increased levels of resting activity relative to those of less-frequent words. Such representations would reach their activation threshold sooner than a lexical representation with a lower level of resting activity. In the latter case, more matching is needed between the visual stimulus and its lexical representation to bring it to threshold, and requiring more time to identify the word stimulus. Secondly, it has also been shown that words with many orthographic neighbors produce a larger negativity than words with few neighbors (Holcomb et al., 2002). These word stimuli generate an increased global lexical activity, increasing the difficulty of integrating the visual information into a specific word meaning. Furthermore, Laszlo and Federmeier (2011) showed that orthographic neighborhood size is a strong predictor of the N400 amplitude. Using a regression analysis of single item ERPs, they showed that N400 amplitude increases for stimuli with many orthographic neighbors as well as for stimuli with higher frequency neighbors. Altogether, these data support the hypothesis that words with many neighbors produce more activation at the lexical level than words with few neighbors. Thus, the N400 component can be seen as reflecting the stabilization of activation at the lexical level and also the mapping of lexical representations onto meaning. Effects of orthographic neighbor primes obtained in the present investigation, between 400 and 600 ms post-target onset, seem to reflect the lexical processing of the target stimulus which is modulated by the prime identification. Larger negativities were found for word neighbor primes (compared to unrelated word primes), highlighting the greater effort required during target processing. According to Colombo (1986), because there is conscious identification of the prime, high-frequency neighbors represent strong competitors for the stimulus word during the selection process in prime recognition and therefore interfere in this process. During prime word identification, selection processes operating to isolate the prime from activated lexical representations

45

must suppress any higher frequency neighbors (Segui and Grainger, 1990). Consequently, for target identification, the lexical representation of the prime has to be strongly inhibited. Thus, inhibitory effects from word neighbor primes can be seen as a side effect of the resolution of competition required for prime identification (Colombo, 1986; Segui and Grainger, 1990), due to transient variations in the activation level of representations at the lexical level. In fact, prime identification modifies the activation (and so inhibition) of some lexical representations, but more critically it leads to a strong activation of its own lexical representation. Furthermore, the more the lexical representation is activated by the prime, the more time it will take for this representation to fully return to its resting state (Versace and Nevers, 2003). Given the assumption that prime activation does not fully return to baseline within 200 ms, the processing of the target stimulus will be affected by the lexical representation of the prime. The notable absence of effects from pseudoword neighbor primes in the 400–600 ms time-window strongly support the account for competition between lexical representations activated by both prime and target stimuli. Pseudoword primes initially activate pre-lexical representations, which in turn leads to the partial activation of lexico-semantic representations of their word neighbors (Andrews, 1997; Grainger and Jacobs, 1996; Holcomb et al., 2002; Sears et al., 1999). Although there is not a full match between the visual input and one lexical representation, identification of pseudoword primes produces activation at the lexical level which is weaker than when the stimulus is a word, because no one lexical representation would tend to dominate. Thus the activation of these lexical representations still leads to competition among them but to a lesser degree than in the case of a word neighbor prime. This is reflected in the present study by a null effect from pseudoword neighbor primes as compared to unrelated pseudoword primes in the 400–600 ms time-window. However, recent data from Burt (2009) showed that pseudoword neighbor primes can slow down target identification (see also Versace, 1998, for similar results), as also shown in the present behavioral data. She argued that these inhibitory effects on target identification are due to lexical representations of neighbors. Shared neighbors are assumed to influence target identification, because their lexical representations have been activated by the prime and re-activated by the target stimulus. In our investigation, it seems unlikely that the pattern of priming effects in the later time-window reflects the activation of lexical representations of neighbors shared by the prime and target stimuli. First, in the present study, the word targets were the same across all experimental conditions. Second, the size of the shared neighborhood was larger for pseudoword-word pairs than for word-word pairs (sizes of shared neighborhood for word-word pairs: mean: 1.52 (SD¼2.19); pseudoword-word pairs: mean: 2.50 (SD¼ 2.20), po.001). According to Burt (2009), we should have observed even stronger effects in the pseudoword neighbor priming condition than in the word priming condition. That was not the case. Even if longer reaction times and more errors were observed for word and pseudoword neighbor priming, a significant increase in the amplitude was only found in the word priming condition in the 400–600 ms

46


time-window. Therefore, the shared neighborhood may certainly play a role during target processing, and the present results demonstrated that, on top of that, the activation of the lexical representation of the prime itself strongly affects target identification. The results of the present investigation provide highly constraining findings with important theoretical consequences. Indeed, this specific pattern demonstrates a direct consequence of prime recognition on target identification processes. In the 400–600 ms time-window, the increasing difficulty seen for target recognition is due to a general competition between lexical representations activated by the target (and the prime). More critical is that, on top of this competition mechanism, the lexical status of the neighbor prime strongly modulates lexical processing of the target stimulus. Conversely, under masking conditions, low-frequency word primes can only produce activation at pre-lexical and lexical levels of representation, due to limited processing time of the prime. Therefore, masked priming effects are automatic and short-lived effects that occur at the initial stages of visual word recognition (Forster et al., 1987; Lukatela and Turvey, 1994; Segui and Grainger, 1990; Sereno, 1991). With unmasked priming, increasing the processing time of the prime allows activation and integration of higher abstract information regarding the prime stimulus. Therefore, such priming effects can reflect of higher level influences, due to the activation of lexico-semantic information. Following the interactive-activation approach, the results of the current study have shown the processing dynamic between pre-lexical (i.e., letter) and lexical level representations. Lexical representations activated by the prime are reactivated by the target presentation through activation from pre-lexical representations to lexical representations and also through activation feedback from activated lexical representations to orthographic pre-lexical representations. Critically, these results have shown that priming effects from orthographically related primes cannot be explained by just an overall activation of lexical representations of word neighbors, but rather that the frequency of the prime plays a major role during target identification processes.

5.

Experimental procedure

5.1.

Experiment 1: method

5.1.1.

Participants

39 participants were recruited for this experiment (23 women, mean age¼ 23 years, SD¼3.67) and were paid for their participation. All of them were native Spanish speakers, with no history of neurological or psychiatric impairment, and with normal or corrected-to-normal vision. None of them reported any reading deficit.

5.1.2.

Design and stimuli

The critical stimuli for this experiment were 228 pairs of letterstrings of 4–6 characters. The first member of each pair was referred to as the prime and the second as the target. All the targets were Spanish words and had a printed frequency higher

than 50 occurrences per million (B-Pal database; Davis and Perea, 2005). All targets had a least one lower frequency orthographic neighbor (average printed frequency¼ 6 occurrences per million). Targets had on average 5.82 orthographic neighbors (SD¼5.32). There were 76 trials in each of three different priming conditions. In the first condition, the target was identical to the prime (e.g., boca-BOCA [mouth-MOUTH]). The second condition was used to test for priming from a lowfrequency neighbor (e.g., beca-BOCA [grant-MOUTH]) and the last condition was used to test for priming from a pseudoword orthographic neighbor (e.g., bica-BOCA [MOUTH]). Orthographically related pseudowords were created by changing a single letter in the target word with a different letter to produce an orthographically legal letter string (e.g., bica-BOCA). They had on average 4.79 orthographic neighbors (SD¼3.63). These three sets of targets were separated into two subtests to create six lists of experimental stimuli presented to different participants. In each of the six lists, there were 38 trials where the target was the same word as the prime (e.g., boca-BOCA), 38 trials where the target was preceded by a word neighbor (e.g., beca-BOCA), 38 trials where the target was preceded by a pseudoword neighbor (e.g., bica-BOCA), and 114 trials where the target was completely unrelated to the prime (e.g., arte-BOCA [skill-MOUTH]; arce-BOCA [mapleMOUTH]; arpe-BOCA [MOUTH]). Across lists and participants, critical targets appeared once in each of the six conditions, and within lists each target stimulus was presented once. In this way, participants saw each target only once but were tested in each experimental condition with different targets. However, across participants each item occurred an equal number of times in both related and unrelated conditions. Type-of-Prime (Identity, Word Neighbor, Pseudoword Neighbor) was crossed with Relatedness (Related, Unrelated) in a 3 2 factorial design. Unrelated prime-target pairs were formed by re-arranging the related prime-target pairs ensuring that there was minimal orthographic overlap and semantic overlap between primes and targets in the re-pairings. Note that this implies that the unrelated primes were lowfrequency words in the word neighbor condition and highfrequency words in the identity condition. Thus, all targets were high-frequency and were preceded either by a highfrequency prime (for testing the identity condition), by a lowfrequency prime (for testing the orthographic word neighbor condition), or by a pseudoword (for testing the pseudoword neighbor condition). In order to avoid participants developing predictive strategies during this experiment, 380 filler pairs of items were added. In order to keep the same proportion of low- and highfrequency words as primes, 76 pairs formed by an unrelated low-frequency prime (average printed frequency¼5 occurrences per million) followed by a low-frequency word target (average printed frequency¼5 occurrences per million) and 76 pairs formed by an unrelated high-frequency word prime (average printed frequency¼105 occurrences per million) followed by a low-frequency target (average printed frequency¼ 5 occurrences per million) were added. Furthermore, in order to have half of the primes as words and the other half as pseudowords, we also added 228 pairs, with 76 pairs of unrelated pseudoword primes followed by a high-frequency target (average printed frequency¼ 128 occurrences per

47


Table 3 – Mean microvolt values and standard deviation at the Midline and at the four groups of electrodes: Left Anterior (average activity of FP1, F3, F7, FC5 and FC1), Right Anterior (FP2, F4, F8, FC6 and FC2), Left Posterior (O1, P3, P7, CP5 and CP1), Right Posterior (O2, P4, P8, CP6 and CP2) for the type of prime and relatedness factors. Identity

Word neighbor

Related

200–400 ms Midline Left Anterior Right Anterior Left Posterior Right Posterior 400–600 ms Midline Left Anterior Right Anterior Left Posterior Right Posterior

Unrelated

Related

Unrelated

Related

Unrelated

M

SD

M

SD

M

SD

M

SD

M

SD

M

SD

3.20 3.19 3.93 .97 2.21

4.73 4.13 3.94 3.46 3.47

2.34 2.85 2.78 1.16 1.92

4.35 3.65 3.46 3.20 3.07

1.58 1.78 2.26 .34 1.37

4.87 4.04 3.86 3.63 3.75

1.46 1.83 1.86 .42 1.18

4.07 3.61 3.34 2.86 2.79

1.69 2.25 2.57 .11 1.31

4.57 3.74 3.65 3.19 3.13

1.32 2.04 2.09 .39 1.12

4.47 3.97 3.64 3.20 3.10

7.10 4.79 6.05 5.01 5.86

5.56 4.71 5.09 3.54 3.83

4.04 2.78 3.72 2.72 3.69

5.53 4.25 4.51 3.86 3.85

3.44 1.73 3.24 1.84 3.30

5.89 4.87 4.67 4.18 4.17

4.10 2.26 3.09 2.82 3.74

5.19 4.16 4.39 3.35 3.60

4.57 2.77 3.98 2.55 3.93

5.50 4.41 4.54 3.93 3.82

4.07 2.46 3.33 2.69 3.62

5.51 4.59 4.40 3.95 3.47

million) and 152 pairs of unrelated pseudowords followed by a low-frequency target (average printed frequency¼5 occurrences per million). All of these pseudowords were created by changing a single letter in a Spanish word to a different letter to produce an orthographically legal letter string (e.g., flira-FLORA [FLORA]). Consequently, half of the total number of trials involved a low-frequency word in target position and the remaining trials involved a high-frequency word in target position. Moreover, the lexical status of the prime did not provide any cue to the lexical status of the target stimulus. Finally, for the purpose of the lexical decision task, 608 pairs of prime – pseudoword targets were added, keeping the exact same proportion of trials involving low-frequency primes, highfrequency primes and pseudoword primes.

5.1.3.

Pseudoword neighbor

Procedure

The presentation of the stimuli and recording of the responses were carried out using Presentation software. All stimuli were presented as white letters centered vertically and horizontally on a black background on a CRT monitor. Participants were informed that two letter strings were going to be subsequently displayed and that they have to make a decision only on the second one. All stimuli were presented in white Courier New font (size 16 pt.). Each trial began with the presentation of a centered fixation cross (þ) which remained on screen for 500 ms. This fixation cross was replaced at the same location on the screen by a lowercase letter-string (i.e., prime stimulus) for 200 ms. The prime was immediately followed by a second letter-string (i.e., target stimulus) in uppercase letters for a duration of 500 ms. The inter-trial interval had a duration of 1000 ms. Participants were instructed to decide as rapidly and as accurately as possible whether or not the target stimulus was a word in Spanish (lexical decision task). They responded yes by pressing the “L” button on the keyboard and no by pressing the “S” button. Reaction times, measured from target onset until participants' response, were accurate to the nearest millisecond. A short practice session was administered before the

main experiment to familiarize participants with the procedure of the lexical decision task. The experiment required approximately 50 min to be completed and included 5 pauses.

5.2.

Experiment 2: method

5.2.1.

Participants

A total of 40 participants were recruited for this experiment (23 women, mean age¼22 years, SD¼ 2.2) and were paid for their participation. All participants were right-handed, with no history of neurological or psychiatric impairment, and with normal or corrected-to-normal vision. None of them reported any reading deficit. Each participant signed an informed consent form before the experiment and was appropriately informed regarding the basic procedure of the experiment, according to the ethical commitments established by the Ethics Committee that approved the experiment. 7 of these participants were excluded from analysis due to an excessive number of artifacts during the experiment or technical issues.

5.2.2.

Design and stimuli

The critical stimuli were 228 pairs of 4–6 letter strings. All the pairs of items were the same as those used in the previous experiment. Thus, in each condition, 38 primes were orthographically related to the word target and 38 primes were totally unrelated to the word target. Across lists and participants, critical targets appeared once in each of the six conditions, and within lists each target stimulus was presented once. In this way, participants saw each target only once but were tested in each experimental condition with different targets. However, across participants each item occurred an equal number of times in both related and unrelated conditions. Type-of-Prime (Identity, Word Neighbor, Pseudoword Neighbor) was crossed with Relatedness (Related, Unrelated) in a 3 2 factorial design.

48


Table 4 – All F-values from a time-course analysis showing main priming effects and interactions of the factor with the Anterior–Posterior (AP) and Hemisphere (H) as distributional factors, for the three types of priming, at consecutive 50 ms epochs from 150 to 600 ms post-target onset.

Identity Identity H Identity AP Word Neighbor Word Neighbor H Word Neighbor AP Pseudoword Neighbor Pseudoword Neighbor H Pseudoword Neighbor AP n

150– 200 ms

200– 250 ms

250– 300 ms

300– 350 ms

350– 400 ms

400– 450 ms

1.08 .68 19.63 .05 .08 2.41 2.36

.18 14.03 1.45 o.01 .02 .28 3.45

.15 10.86 7.29 .19 4.25 .20 .08

1.46 9.96 13.65 .11 5.48 .47 .20

15.6nn 13.71 nn .55 1.17 8.61 n .03 .03

46.27 2.64 4.11 .90 10.33 1.08 6.34

nn

.08 8.43

n

nn

n n

n

2.40

1.41

2.29

.44

.56

7.16

n n

n

9.24 n

2.46

n

6.94 .01

nn

n

n

n

450– 500 ms 48.31 .65 1.77 1.09 12.16 .80 .23

nn

14.51

n

2.96

n

500– 550 ms 23.51 .01 o.01 6.59 6.95 8.74 o.01

nn

n n n

550– 600 ms 13.61 2.22 6.29 2.62 5.53 6.29 o.01

1.72

.01

2.31

2.01

n

n

n n

po.05. po.01.

nn

5.2.3.

Procedure

The procedure for stimulus presentation was identical to that used in Experiment 1. Participants were asked to refrain from blinking and moving their eyes when the fixation stimulus appeared on the screen to minimize eye blink artifact during the recorded trials. The experiment required approximately 50 min to be complete and included 5 pauses.

5.2.4.

Electroencephalogram recording procedure

After completing informed consent, participants were seated in a comfortable chair in a sound-attenuated and dimly illuminated room. The electroencephalogram (EEG) was recorded continuously through a 32-channels Brain-Amp system from 27 Ag/AgCl electrodes mounted on an elastic cap (Easy Cap) that was positioned according the 10–10 International system (Fp1/Fp2, F3/F4, F7/F8, FC1/FC2, FC5/ FC6, C3/C4, T7/T8, CP1/CP2, CP5/CP6, P3/P4, P7/P8, O1/O2, Fz, Cz, Pz). The montage included 3 midline sites and 12 sites over each hemisphere. Four additional electrodes were used to monitor eye movements and blinks (two placed at lateral canthi and two below the eyes). An additional electrode placed over the left mastoid (A1) was used as online reference and a final electrode placed over the right mastoid (A2). For all scalp electrodes impedances were maintained below 5 kΩ, and below 10 kΩ for EOG electrodes (electrooculography). Continuous EEG was digitized at 250 Hz and filtered offline (High-pass: .01 Hz, 12 dB/octave; Low-pass: 30 Hz, 48 dB/ octave) using Brain Analyzer Software. All electrode sites were re-referenced offline to the average activity of the two mastoids. Epochs with eye movements, blinks, or electrical activities greater than þ/ 80 μV were rejected. To maintain an acceptable signal-to-noise ratio, a lower limit of 30 artifact-free trials per participant per condition was set. On this basis, 7 participants were excluded from further analysis.

5.2.5.

Data analysis

ERPs were calculated by averaging the EEG time-locked to a point 200 ms pre-target onset and lasting until 700 ms posttarget onset. A 200 ms pre-target period was used as the

baseline3. Only trials without muscle artifact or eye movement/blink activity were included in the averaging process (11.12% of rejected trials). These resulted in a highly-similar amount of artifact-free segments across conditions (Identity: 92.43%; Unrelated word: 90.12%; Word neighbor: 86.37%; Unrelated word: 90.12%; Nonword neighbor: 86.37%; Unrelated nonword: 87.88%). Then these artifact-free and errorfree segments were averaged and analyzed only. Following visual inspection and based on the previous studies using the priming paradigm (i.e., Chauncey et al., 2008; Holcomb and Grainger, 2007; Vergara-Martínez et al., 2011), analyses on average amplitudes were run in two time intervals: 200–400 ms and 400–600 ms. In both time intervals we ran two distinct analyses of variance. In all cases the Greenhouse–Geisser correction was applied in the case of lack of sphericity in the data (Geisser and Greenhouse, 1959). The first Midline ANOVA was conducted on the midline electrodes: an initial three-way ANOVA was run crossing the Electrode factor (three levels: Fz, Cz, Pz) with the Typeof-Prime factor (three levels: Identity, Word Neighbor, Pseudoword Neighbor) and with the Relatedness factor (two levels: Related, Unrelated). In order to evaluate possible lateralized ERP effects we also conducted a second ANOVA on the remaining electrodes. We grouped the activity elicited by contiguous electrodes calculating the mean values in four homogeneously distributed lateralized groups: Left Anterior group (FP1, F3, F7, FC5 and FC1), Right Anterior group (FP2, F4, F8, FC6 and FC2), Left Posterior group (O1, P3, P7, CP5 and CP1), and Right Posterior group (O2, P4, P8, CP6 and CP2). We employed an approach to data analysis by which the head is divided up into four groups of electrode sites. This approach allows for the analysis of the topographical distribution of the effects, because it provides a thorough analysis of the entire 3 Note: another analysis of the ERP data with a 200 ms preprime period as baseline was also performed. By doing so, we ensured that the ERP components and effects were similar based on visual inspection of the grand averages as those reported here with a 200 ms pre-target period as baseline.


head and at the same time allows for the observation of small regional effects (see Carreiras et al., 2009a, 2009b; Molinaro and Carreiras, 2010, for a similar approach). Mean amplitudes in each time window entered a four-way overall ANOVA with two topographical factors, Anterior–Posterior (two levels: Anterior, Posterior) and Hemisphere (two levels: Left, Right), the three level Type-of-Prime factor and the two level Relatedness factor. Relevant main effects or interactions involving the experimental factors were further evaluated by ANOVAs with the same design, comparing the conditions in a pairwise manner. Table 3 provides mean voltage values per experimental condition at the Midline and at the four groups of Electrodes for each epoch used in the ANOVAs.

Acknowledgments This research has been partially supported by Postdoctoral Fellowship from the Fyssen Foundation (Paris) and the European Commission (Marie Curie fellowship, FP7-PEOPLE2011-IEF, Project number 301901, attributed to Stéphanie Massol), by Grants PSI2012-32350 and PSI2012-31448 from the Spanish Ministry of Economy and Competitiveness and by Grant CONSOLIDER-INGENIO2010 CSD2008-00048 from the Spanish Government and ERC-2011-ADG-295362 Grant from the European Research Council.

r e f e r e n c e s

Andrews, S., 1989. Frequency and neighborhood size effects on lexical access: activation or search? J. Exp. Psychol.: Learn. Mem. Cognit. 15, 802–814, http://dx.doi.org/10.1037/02787393.15.5.802. Andrews, S., 1992. Frequency and neighborhood effects on lexical access: lexical similarity or orthographic redundancy? J. Exp. Psychol.: Learn. Mem. Cognit. 18, 234–254, http://dx.doi.org/ 10.1037/0278-7393.18.2.234. Andrews, S., 1997. The effect of orthographic similarity on lexical retrieval: resolving neighborhood conflicts. Psychon. Bull. Rev. 4, 439–461, http://dx.doi.org/10.3758/bf03214334. Balota, D.A., Yap, M.J., Cortese, M.I, Watson, J.M., 2008. Beyond response latency: an RT distributional analysis of semantic priming. J. Mem. Lang. 59, 495–523, http://dx.doi.org/10.1016/j. jml.2007.10.004. Barber, H., Vergara, M., Carreiras, M., 2004. Syllabe-frequency effects in visual word recognition: evidence from ERPs. NeuroReport 15, 545–548, http://dx.doi.org/10.1097/00001756200403010-00032. Bentin, S., McCarthy, G., Wood, C.C., 1985. Event-related potentials, lexical decision, and semantic priming. Electroencephalogr. Clin. Neurophysiol. 60, 343–355, http://dx. doi.org/10.1016/0013-4694(85)90008-2. Bentin, S., Mouchetant-Rostaing, Y., Giard, M.H., Echallier, J.F., Pernier, J., 1999. ERP manifestations of processing printed words at different psycholinguistic levels: time-course and scalp distribution. J. Cogn. Neurosci. 11, 35–60, http://dx.doi. org/10.1162/089892999563373. Bodner, G.E., Masson, M.E.J., 1997. Masked repetition priming of words and nonwords: evidence for a nonlexical basis for priming. J. Mem. Lang. 37, 268–293. Bodner, G.E., Masson, M.E.J., 2003. Beyond spreading activation: an influence of relatedness proportion on masked semantic

49

priming. Psychon. Bull. Rev. 10, 645–652, http://dx.doi.org/ 10.3758/bf03196527. Bowers, J.S., 2000. In defense of abstractionist theories of repetition priming and word identification. Psychon. Bull. Rev. 7, 83–99, http://dx.doi.org/10.3758/BF03210726. Bowers, J.S., 2003. An abstractionist account of masked and longterm priming. In: Kinoshita, S., Lupker, S.J. (Eds.), Masked Priming: The State of the Art. Psychology Press, New York, pp. 39–55. Burt, J.S., 2009. Identifiable orthographically similar word primes interfere in visual word identification. J. Mem. Lang. 61 (3), 259–284, http://dx.doi.org/10.1016/j.jml.2009.06.003. Carreiras, M., Dunãbeitia, J.A., Molinaro, N., 2009a. Consonants and vowels contribute differently to visual word recognition: ERPs of relative position priming. Cereb. Cortex 19, 2659–2670, http://dx.doi.org/10.1093/cercor/bhp019. Carreiras, M., Perea, M., Grainger, J., 1997. Effects of orthographic neighborhood in visual word recognition: cross-task comparisons. J. Exp. Psychol.: Learn. Mem. Cognit. 23, 857–871, http://dx.doi.org/10.1037/0278-7393.23.4.857. Carreiras, M., Perea, M., Vergara, M., Pollatsek, A., 2009b. The time course of orthography and phonology: ERP correlates of masked priming effects in Spanish. Psychophysiology 46, 1113–1122, http://dx.doi.org/10.1111/j.1469-8986.2009.00844. Chauncey, K., Holcomb, P.J., Grainger, J., 2008. Effects of stimulus font and size on masked repetition priming: an ERP investigation. Lang. Cogn. Process. 23 (1), 183–200. Coch, D., Mitra, P., 2010. Word and pseudoword superiority effects reflected in the ERP waveform. Brain Res. 1329, 159–174, http: //dx.doi.org/10.1016/j.brainres.2010.02.084. Colombo, L., 1986. Activation and inhibition with orthographically similar words. J. Exp. Psychol.: Hum. Percept. Perform. 12, 226–234, http://dx.doi.org/10.1037/00961523.12.2.226. Coltheart, M., Davelaar, E., Jonasson, J.T., Besner, D., 1977. Access to the internal lexicon. In: In: Dornic, S. (Ed.), Attention and Performance, IV. Erlbaum, Hillsdale, NJ, pp. 535–555, http://dx. doi.org/10.3758/bf03197471. Davis, C.J., Lupker, S.J., 2006. Masked inhibitory priming in English: evidence for lexical inhibition. J. Exp. Psychol.: Hum. Percept. Perform. 32, 668–687, http://dx.doi.org/10.1037/00961523.32.3.668. Davis, C.J., Perea, M., 2005. BuscaPalabras: a program for deriving orthographic and phonological neighborhood statistics and other psycholinguistic indices in Spanish. Behav. Res. Methods 37, 665–671, http://dx.doi.org/10.3758/bf03192738. De Moor, W., Brysbaert, M., 2000. Neighbourhood-frequency effects when primes and targets have different lengths. Psychol. Res. 63, 159–162, http://dx.doi.org/10.1007/ PL00008174. De Moor, W., Verguts, T., 2006. When are neighbours hostile? Inhibitory neighbour effects in visual word recognition. Psychol. Res. 70, 359–366, http://dx.doi.org/10.1007/s00426005-0225. Dehaene, S., Naccache, L., Cohen, L., LeBihan, D., Mangin, J.F., Poline, J.B., Rivie`re, D., 2001. Cerebral mechanisms of word masking and unconscious repetition priming. Nat. Neurosci. 4, 752–758, http://dx.doi.org/10.1038/89551. Durgunoglu, A.Y., Neely, J.H., 1987. On obtaining episodic priming in a lexical decision task following paired-associate learning. J. Exp. Psychol.: Learn. Mem. Cognit. 13, 206–222, http://dx.doi. org/10.1037/0278-7393.13.2.206. Forbach, G.B., Stanners, R.F., Hochaus, L., 1974. Repetition and practice effects in a lexical decision task. Mem. Cognit. 2 (2), 337–339, http://dx.doi.org/10.3758/BF03209005. Forster, K.I., 1987. Form-priming with masked primes: the bestmatch hypothesis. In: In: Coltheart, M. (Ed.), Attention & Performance, XII. Erlbaum, Hillsdale, N.J., pp. 127–146.

50


Forster, K., Booker, J., Schacter, D.L., Davis, C., 1990. Masked repetition priming: lexical activation or novel memory trace? Bull. Psychon. Soc. 28, 341–345, http://dx.doi.org/10.3758/ bf03334039. Forster, K.I., Davis, C.W., 1984. Repetition priming and frequency attenuation in lexical access. J. Exp. Psychol.: Learn. Mem. Cognit. 10, 680–698, http://dx.doi.org/10.1037/02787393.10.4.680. Forster, K.I., Davis, C.W., Schoknecht, C., Carter, R., 1987. Masked priming with graphemically related forms: repetition or partial activation?. Q. J. Exp. Psychol. 39, 211–251, http://dx. doi.org/10.1080/14640748708401785. Forster, K.I., Mohan, K., Hector, J., 2003. The mechanics of masked priming. In: Kinoshita, S., Lupker, S.J. (Eds.), Masked Priming: State of the Art. Psychology Press, Hove, UK, pp. 3–37 http://dx.doi.org/10.4324/9780203502846. Forster, K.I., Veres, C., 1998. The prime lexicality effect: formpriming as a function of prime awareness, lexical status, and discrimination difficulty. J. Exp. Psychol.: Learn. Mem. Cognit. 24, 498–514, http://dx.doi.org/10.1037/02787393.24.2.498. Frost, R., Forster, K.I., Deutsch, A., 1997. What can we learn from the morphology of Hebrew? A masked priming investigation of morphological representation. J. Exp. Psychol.: Learn. Mem. Cognit. 23, 829–856, http://dx.doi.org/10.1037/02787393.23.4.829. Frost, R., Katz, L., Bentin, S., 1987. Strategies for visual word recognition and orthographical depth: a multilingual comparison. J. Exp. Psychol.: Hum. Percept. Perfom. 13, 104–115. Geisser, S., Greenhouse, S., 1959. On methods in the analysis of profile data. Psychometrika 24, 95–112, http://dx.doi.org/ 10.1007/bf02289823. Gobin, P., Faı¨ta-Ainseba, F., Mathey, S., 2012. Orthographic priming also depends on the emotional valence of the neighbor and prime duration: an ERP study. J. Neurolinguist. 25 (3), 178–193, http://dx.doi.org/10.1016/j. jneuroling.2011.12.001. Gobin, P., Mathey, S., 2010. The influence of emotional orthographic neighbourhood in visual word recognition. Curr. Psychol. Lett. 26, 1–11, http://dx.doi.org/10.1037/e636952013223. Gomez, P., Perea, M., Ratcliff, R., 2013. A diffusion model account of masked vs. unmasked priming: are they qualitatively different?. J. Exp. Psychol.: Hum. Percept. Perform. 39, 1731–1740, http://dx.doi.org/10.1037/a0032333. Grainger, J., 1990. Word frequency and neighborhood frequency effects in lexical decision and naming. J. Mem. Lang. 29, 228–244, http://dx.doi.org/10.1016/0749-596x(90)90074-a. Grainger, J., 2008. Cracking the orthographic code: an introduction. Lang. Cogn. Process. 23 (1), 1–35, http://dx.doi. org/10.1080/01690960701578013. Grainger, J., Ferrand, L., 1996. Masked orthographic and phonological priming in visual word recognition and naming: cross-task comparisons. J. Mem. Lang. 35, 623–647, http://dx. doi.org/10.1006/jmla.1996.0033. Grainger, J., Holcomb, P.J., 2009. Watching the Word Go by: on the Time-course of Component Processes in Visual Word Recognition. Lang. Linguist. Compass 3 (1), 128–156, http://dx. doi.org/10.1111/j.1749-818x.2008.00121. Grainger, J., Jacobs, A.M., 1996. Orthographic processing in visual word recognition: a multiple read-out model. Psychol. Rev. 103, 518–565, http://dx.doi.org/10.1037/0033-295x.103.3.518. Grainger, J., Jacobs, A.M., 1993. Masked partial-word priming in visual word recognition: effects of positional letter frequency. J. Exp. Psychol.: Hum. Percept. Perform. 19, 951–964. Grainger, J., Kiyonaga, K., Holcomb, P.J., 2006. The Time Course of Orthographic and Phonological Code Activation. Psychol. Sci.

17 (12), 1021–1026, http://dx.doi.org/10.1111/j.14679280.2006.01821. Grainger, J., Lopez, D., Eddy, M., Dufau, S., Holcomb, P.J., 2012. How word frequency modulates masked repetition priming: an ERP investigation. Psychophysiology 49, 604–616, http://dx.doi.org/ 10.1111/j.1469-8986.2011.01337. Grainger, J., O’Regan, K., Jacobs, A., Segui, J., 1989. On the role of competing word units in visual word recognition: the neighborhood frequency effect. Percept. Psychophys. 45, 189–195, http://dx.doi.org/10.3758/bf03210696. Grainger, J., Segui, J., 1990. Neighborhood frequency effects in visual word recognition: a comparison of lexical decision and masked identification latencies. Percept. Psychophys. 47, 191–198, http://dx.doi.org/10.3758/bf03205983. Holcomb, P.J., Grainger, J., 2006. On the time-course of visual word recognition: an ERP investigation using masked repetition priming. J. Cogn. Neurosci. 18, 1631–1643, http://dx.doi.org/ 10.1162/jocn.2006.18.10.1631. Holcomb, P.J., Grainger, J., 2007. Exploring the temporal dynamics of visual word recognition in the masked repetition priming paradigm using event-related potentials. Brain Res. 1180, 39–58. Holcomb, P.J., Grainger, J., O’Rourke, T., 2002. An electrophysiological study of the effects of orthographic neighborhood size on printed word perception. J. Cogn. Neurosci. 14, 938–950, http://dx.doi.org/10.1162/ 089892902760191153. Holcomb, P.J., Reder, L., Misra, M., Grainger, J., 2005. The effects of prime visibility on ERP measures of masked priming. Cogn. Brain Res. 24, 155–172. Humphreys, G.W., Besner, D., Quinlan, P.T., 1988. Event perception and the word repetition effect. J. Exp. Psychol.: Gen. 117, 51–67, http://dx.doi.org/10.1037/0096-1523.14.2.203. Jacobs, A.M., Grainger, J., 1992. Testing a semistochastic variant of the interactive activation model in different word recognition experiments. J. Exp. Psychol.: Hum. Percept. Perform. 18, 1174–1188, http://dx.doi.org/10.1037/0096-1523.18.4.1174. Kinoshita, S., Lupker, S.J. (Eds.), 2003. Masked Priming: The State of the Art. Psychology Press, New York (DOI: 10.4324/ 9780203502846). Kinoshita, S., Mozer, M.C., Forster, K.I., 2011. Dynamic adaptation to history of trial difficulty explains the effect of congruency proportion on masked priming. J. Exp. Psychol.: Gen. 40, 622–636 (http://dx.doi.org/10.1016/j.cognition.2007.11.011). Kutas, M., Hillyard, S.A., 1980. Reading senseless sentences: brain potentials reflect semantic incongruity. Science 207, 203–205, http://dx.doi.org/10.1126/science.7350657. Laszlo, S., Federmeier, K.D., 2011. The N400 as a snapshot of interactive processing: evidence from regression analyses of orthographic neighbor and lexical associate effects. Psychophysiology 48, 176–186, http://dx.doi.org/10.1111/ j.1469-8986.2010.01058. Laszlo, S., Federmeier, K.D., 2014. Never seem to find the time: evaluating the physiological time course of visual word recognition with regression analysis of single-item eventrelated potentials. Lang. Cognit. Neurosci. 29 (5), 642–661, http: //dx.doi.org/10.1080/01690965.2013.866259. Lukatela, G., Turvey, M.T., 1994. Visual lexical access is initially phonological: 1. Evidence from associative priming by words, homophones, and pseudohomophones. J. Exp. Psychol.: Gen. 123, 107–128, http://dx.doi.org/10.1037/0096-1523.17.4.951. Massol, S., 2012. Effets d’amorçage masque´ par voisinage orthographique : Etudes Electrophysiologiques. L’Anneé Psychol. 112, 361–402, http://dx.doi.org/10.4074/ s000350331200303x. Massol, S., Grainger, J., Dufau, S., Holcomb, P.J., 2010. Masked priming from Orthographic Neighbors: an ERP Investigation.


J. Exp. Psychol.: Hum. Percept. Perform. 36, 162–174, http://dx. doi.org/10.1037/a0017614. Massol, S., Grainger, J., Midgley, K.J., Holcomb, P.J., 2012. Masked repetition priming of letter-in-string identification: an ERP investigation. Brain Res. 1472, 74–88, http://dx.doi.org/10.1016/ j.brainres.2012.07.018. Massol, S., Midgley, K.J., Holcomb, P.J., Grainger, J., 2011. When less is more: feedback, priming, and the pseudoword superiority effect. Brain Res. 1386, 153–164, http://dx.doi.org/ 10.1016/j.brainres.2011.02.050. McCandliss, B.D., Cohen, L., Dehaene, S., 2003. The visual word form area: expertise for reading in the fusiform gyrus. Trends Cogn. Sci. 13, 155–161, http://dx.doi.org/10.1016/s1364-6613(03) 00134-7. McClelland, J.L., Rumelhart, D.E., 1981. An interactive activation model of context effects in letter perception: Part 1. An account of basic findings. Psychol. Rev. 88, 375–407, http://dx. doi.org/10.1037/0033-295x.88.5.375. Midgley, K.J., Holcomb, P.J., Grainger, J., 2009. Masked repetition and translation priming in second language learners: a window on the time-course of form and meaning activation using ERPs. Psychophysiology 46, 551–565, http://dx.doi.org/ 10.1111/j.1469-8986.2009.00784.x. Molinaro, N., Carreiras, M., 2010. Electrophysiological evidence of interaction between contextual expectation and semantic integration during the processing of collocations. Biol. Psychol. 83, 176–190, http://dx.doi.org/10.1016/j. biopsycho.2009.12.006. Monsell, S., 1985. Repetition and the lexicon. In: Ellis, A.W. (Ed.), Progress in the Psychology of Language, vol. 2. Erlbaum, London, pp. 147–195. Nakayama, M., Sears, C.R., Lupker, S.J., 2008. Masked priming with orthographic neighbors: a test of the lexical competition assumption. J. Exp. Psychol.: Hum. Percept. Perform. 34, 1236–1260, http://dx.doi.org/10.1037/0096-1523.34.5.1236. Neely, J.H., 1991. Semantic priming effects in visual word recognition: a selective review of current findings and theory. In: Besner, D., Humphreys, G.W. (Eds.), Basic Processes in Reading: Visual Word Recognition. Erlbaum, Hillsadale, NJ, pp. 264–336, http://dx.doi.org/10.2307/1423186. Neville, H.J., Mills, D.L., Lawson, D.S., 1992. Fractionating language: different neural subsystems with different sensitive periods. Cereb. Cortex 2, 244–258, http://dx.doi.org/10.1093/ cercor/2.3.244. Perea, M., Carreiras, M., 1998. Effects of syllable frequency and syllable neighborhood frequency in visual word recognition. J. Exp. Psychol.: Hum. Percept. Perform. 24, 134–144, http://dx. doi.org/10.1037/0096-1523.24.1.134. Perea, M., Lupker, S.J., 2003. Transposed-letter confusability effects in masked form priming. In: Kinoshita, S., Lupker, S.J. (Eds.), Masked Priming: State of the Art. Psychology Press, Hove, UK, pp. 97–120. Perea, M., Lupker, S.J., 2004. Can CANISO activate CASINO? Transposed-letter similarity effects with nonadjacent letter positions. J. Mem. Lang. 51, 231–246. Perea, M., Pollatsek, A., 1998. The effects of neighborhood frequency in reading and lexical decision. J. Exp. Psychol.: Hum. Percept. Perform. 24, 767–777, http://dx.doi.org/10.1037/ 0096-1523.24.3.767. Perea, M., Rosa, E., 2000a. Repetition and form priming interact with neighborhood density at a short stimulus-onset asynchrony. Psychon. Bull. Rev. 7, 668–677.

51

Perea, M., Rosa, E., 2000b. The density constraint also occurs with unmasked, visible primes in the lexical decision task. Curr. Psychol. Lett. 2, 61–72. Perry, J., Lupker, S.J., Davis, C., 2008. An evaluation of the interactive-activation model using masked partial-word priming. Lang. Cogn. Process. 23, 36–68, http://dx.doi.org/ 10.1080/01690960701578112. Rajaram, F., Neely, J.H., 1992. Dissociative masked repetition priming and word frequency effects in lexical decision and episodic recognition tasks. J. Mem. Lang. 31, 152–182. Robert, C., Mathey, S., 2007. Aging and lexical inhibition: the orthographic neighborhood frequency effect in young and older adults. J. Gerontol.: Psychol. Sci. 62, 340–342, http://dx. doi.org/10.1093/geronb/62.6.p340. Rugg, M.D., 1990. Event-related brain potentials dissociate repetition effects of high- and low-frequency words. Mem. Cognit. 18, 367–379, http://dx.doi.org/10.3758/bf03197126. Sears, C.R., Hino, Y., Lupker, S.J., 1999. Orthographic neighbourhood effects in parallel distributed processing models. Can. J. Exp. Psychol. 53, 220–230, http://dx.doi.org/ 10.1037/h0087311. Segui, J., Grainger, J., 1990. Priming word recognition with orthographic neighbors: effects of relative prime-target frequency. J. Exp. Psychol.: Hum. Percept. Perform. 16, 65–76, http://dx.doi.org/10.1037/0096-1523.16.1.65. Sereno, J.A., 1991. Graphemic, associative, and syntactic priming effects at a brief stimulus onset asynchrony in lexical decision task and naming. J. Exp. Psychol.: Learn. Mem. Cognit. 17, 459–477, http://dx.doi.org/10.1037/0278-7393.17.3.459. Tarkiainen, A., Helenius, P., Hansen, P.C., Cornelissen, P., Salmelin, R., 1999. Dynamics of letter string perception in the human occipitotemporal cortex. Brain 122, 2119–2131, http: //dx.doi.org/10.1093/brain/122.11.211. Van Petten, C., Kutas, M., 1990. Interactions between sentence context and word frequency in event-related brain potentials. Mem. Cognit. 18 (4), 380–393, http://dx.doi.org/10.3758/ bf03197127. Vergara-Martıńez, M., Perea, M., Marıń, A., Carreiras, M., 2011. The processing of consonants and vowels during letter identity and letter position assignment in visual-word recognition: An ERP study. Brain & Language 118, 105–117, http://dx.doi.org/10.1016/j.bandl.2010.09.006. Versace, R., 1998. Frequency and prime duration effects on repetition priming and orthographic priming with words and pseudowords. Cah. de Psychol. Cogn./Curr. Psychol. Cognit. 17, 535–556. Versace, R., Nevers, B., 2003. Word frequency effect on repetition priming as a function of prime duration and delay between the prime and the target. Br. J. Psychol. 94, 389–408, http://dx. doi.org/10.1348/000712603767876299. Whitlow, J.W., 1990. Differential sensitivity of perceptual identification for words and pseudowords to test expectations: implications for the locus of word frequency effects. J. Exp. Psychol.: Learn. Mem. Cognit. 16, 837–851, http: //dx.doi.org/10.1037/0278-7393.16.5.837. Whitlow, J.W., Cebollero, A., 1989. The nature of word frequency effects on perceptual identification. J. Exp. Psychol.: Learn. Mem. Cognit. 15, 643–656, http://dx.doi.org/10.1037/02787393.15.4.643.

Lexical ambiguity, semantic context, and visual word recognition.

Priming Lexical Neighbors of Spoken Words: Effects of Competition and Inhibition.

Resolving the orthographic ambiguity during visual word recognition in Arabic: an event-related potential investigation.

Lexical support for phonetic perception during nonnative spoken word recognition.

Phonology and orthography in visual word recognition: evidence from masked non-word priming.

Semantic priming of word pronunciation and lexical decision in schizophrenia.

Evaluating cognitive models of visual word recognition using fMRI: Effects of lexical and sublexical variables.

Individual differences in involvement of the visual object recognition system during visual word recognition.

Suprasegmental lexical stress cues in visual speech can guide spoken-word recognition.

Resolving the locus of cAsE aLtErNaTiOn effects in visual word recognition: Evidence from masked priming.

Prelexical facilitation and lexical interference in auditory word recognition.

Grammatical context constrains lexical competition in spoken word recognition.

English Listeners Use Suprasegmental Cues to Lexical Stress Early During Spoken-Word Recognition.

Large-corpus phoneme and word recognition and the generality of lexical context in CVC word perception.

Visual word recognition across the adult lifespan.

Masked priming and ERPs dissociate maturation of orthographic and semantic components of visual word recognition in children.

Syntactic priming during sentence comprehension: evidence for the lexical boost.

Words translated in sentence contexts produce repetition priming in visual word comprehension and spoken word production.

Visual processing: microcircuits unmasked.

Computational Modeling of Morphological Effects in Bangla Visual Word Recognition.

Rapid modulation of spoken word recognition by visual primes.

Does bold emphasis facilitate the process of visual-word recognition?

Real-time functional architecture of visual word recognition.

Evidence for Separate Contributions of High and Low Spatial Frequencies during Visual Word Recognition.