Abstract

Spoken language production involves lexical-semantic access and phonological encoding. A theoretically important question concerns the relative time course of these two cognitive processes. The predominant view has been that semantic and phonological codes are accessed in successive stages. However, recent evidence seems difficult to reconcile with a sequential view but rather suggests that both types of codes are accessed in parallel. Here, we used ERPs combined with the “blocked cyclic naming paradigm” in which items overlapped either semantically or phonologically. Behaviorally, both semantic and phonological overlap caused interference relative to unrelated baseline conditions. Crucially, ERP data demonstrated that the semantic and phonological effects emerged at a similar latency (∼180 msec after picture onset) and within a similar time window (180–380 msec). These findings suggest that access to phonological information takes place at a relatively early stage during spoken planning, largely in parallel with semantic processing.

INTRODUCTION

An important issue in psychology in general, and psycholinguistics in particular, is how multiple cognitive processes take place across time and with regard to one another. Spoken language production involves lexical-semantic and phonological encoding. At the former level, a cohort of semantically related lexical nodes is activated and a lexical target node is selected among co-activated competitors, whereas at the latter level, phonological forms of words are accessed, which enables the following articulation. How cognitive action takes place across semantic and phonological encoding has been one of the hotly debated topics in the domain of language production.

Traditionally, the predominant view held that semantic and phonological codes are processed in two sequential steps, with access to semantic codes preceding phonological processing, and semantic processing is completed before phonological processing. This assumption is embedded in “discrete” or ”serial” models of production, such as those by Garrett (1975) and Levelt, Roelofs, and Meyer (1999). However, various findings from the literature on speech errors, as well as from experimental tasks that elicit spoken responses in the laboratory have cast doubt on the assumption of strict seriality, and hence, a number of formal models of spoken production have introduced some degree of “interactivity” between processing levels, with the most prominent interactive model proposed by Dell (1986; Dell, Schwartz, Martin, Saffran, & Gagnon, 1997). Rapp and Goldrick (2000) provide a detailed theoretical unfolding of the notions of seriality, interactivity, cascadedness, and feedback. Even with interactive principles, most models still adhere to a broad notion of sequentiality, such that, in the preparation of a spoken utterance, semantic codes are accessed first, whereas phonological properties are retrieved later. By contrast, a number of recent studies have reported findings on the basis of which the authors have challenged this prominent sequential view and have instead postulated a parallel account according to which speakers rapidly access semantic and phonological information largely in parallel (e.g., Strijkers, Costa, & Pulvermüller, 2017; Miozzo, Pulvermüller, & Hauk, 2015). An exploration of the relative time course of the two cognitive processes is therefore theoretically important as it would provide a critical test of sequential versus parallel processing. In this study, we investigated the issue by tracking the temporal dynamics of semantic and phonological processes in spoken production via high time-resolution electrophysiological measurement, that is, ERPs.

Early electrophysiological investigations on the time course of semantics and phonology employed covert (rather than overt) naming tasks, and largely based on this evidence, Indefrey and Levelt (2004) provided estimates of temporal windows corresponding to the processes underlying word production. Spoken production was described as a serial succession of processes elicited by a single-word production episode such as in object naming, with semantic processing taking place from 200 to 275 msec, and phonological encoding from 275 to 450 msec after picture onset. However, there is increasing concern that, from a methodological perspective, these early studies were not ideally suited to detect the time course of the underlying processes. For example, studies relying on motor response preparation or inhibition indexed by “lateralized readiness potentials” and N200 in semantic or phonological classification tasks suggested that the availability of semantic information precedes that of phonological information (Schmitt, Münte, & Kutas, 2000; Van Turennout, Hagoort, & Brown, 1997; but see Abdel Rahman & Sommer, 2003). As these early ERP studies did not involve actual overt speech production, but rather relied on complex meta-linguistic decisions linked to button-press responses (e.g., semantic-phonological categorization task in van Turennout et al. 1997), they could reflect response decision rather than on-line timing of cognitive processes underlying a naming response (see Strijkers & Costa, 2011). Moreover, it has been argued that lateralized readiness potentials or N200 are only informative as to the termination of semantic and phonological processes but not their initiation (Laganaro & Perret, 2011; Camen, Morand, & Laganaro, 2010; Abdel Rahman & Sommer, 2003). Finally, the complexity associated with these experimental tasks allows for several alternative interpretations (see Strijkers & Costa, 2016, for a review).

More recent studies have successfully combined overt spoken production with EEG measurements. A particularly rich source of empirical evidence for overt spoken production has been derived from the picture–word interference (PWI) task. Relative to a condition in which target picture and distractor word are unrelated, a semantic relationship slows down naming, whereas a phonological relationship speeds up responses (Starreveld & La Heij, 1995; Schriefers, Meyer, & Levelt, 1990; Glaser & Düngelhoff, 1984). This semantic interference has arguably been assumed to arise at the stage of lexical-semantic retrieval, and phonological facilitation arises at the stage of phonological encoding. The current results clearly demonstrate a statistical interaction between semantic and phonological relatedness in PWI tasks, which were taken as evidence for the nonserial view (Taylor & Burke, 2002; Damian & Martin, 1999; Starreveld & La Heij, 1995, 1996). Dell'Acqua et al. (2010) used ERPs combined with the PWI task to track the time course of semantic and phonological encoding and found sequential time windows for the semantic and phonological effects (see Zhu, Damian, & Zhang, 2015, for the similar findings in Mandarin spoken production). However, concerns about the PWI task for investigating the time course of production have been voiced (Strijkers & Costa, 2011). One possible criticism is that the exact locus of semantic effects from the PWI task remains controversial. For example, semantic effects have been attributed to semantic-lexical level (e.g., Levelt et al., 1999; Starreveld & La Heij, 1996; Roelofs, 1992; Schriefers et al., 1990; Glaser & Glaser, 1982), or, alternatively, a postlexical articulatory level by the “response exclusion” hypothesis (e.g., Mahon, Costa, Peterson, Vargas, & Caramazza, 2007). Moreover, the superimposition of a visually complex distractor could itself delay the cognitive processes associated with object name production (see Qu & Damian, 2020, for a review).

A number of recent studies have reported findings that suggest a parallel time course of semantics and phonology, a pattern that is difficult to reconcile with a sequential model that would predict more sequential time signatures (e.g., Strijkers et al., 2017; Miozzo et al., 2015; Strijkers, Costa, & Thierry, 2010; for reviews, see Munding, Dubarry, & Alario, 2016; Strijkers & Costa, 2016). First, it has been demonstrated that phonological manipulations modulated ERPs starting around 200 msec after picture presentation. For example, Qu, Damian, and Kazanina (2012) asked Chinese speakers to name colored objects with adjective–noun phrases and found that phoneme overlap between adjective and noun-modulated ERPs starting from 200 msec after picture onset (see also Qu, Feng, Hou, & Damian, 2020; Yu, Mo, & Mo, 2014, for a similar finding with various tasks). Such an early onset of phonological effects in spoken production would be incompatible with the time estimates provided by Indefrey and Levelt (2004; Indefrey, 2011) and, in fact, would imply that semantic and phonological processing begins roughly simultaneously. Using a multiple linear regression approach in a magnetoencephalography (MEG) study of picture naming, Miozzo et al. (2015) manipulated variables that were assumed to primarily impact semantic (semantic features and action features) and phonological processing (word length, phonological neighborhood size). They found an early and simultaneous latency of semantic and phonological effects at around 150 msec upon object presentation. In another study, Strijkers et al. (2017) went beyond “adjacent” processing layers (lexical-semantic vs. phonological encoding) and investigated the interactivity of a more extreme contrast targeting the lexical-semantic and phonetic-articulatory processes in object naming. Lexical-semantic processing was assessed by manipulating word frequency, whereas phonetic-articulatory properties were manipulated by the place of articulation of word-initial phonemes (lip vs. tongue, e.g., Monkey vs. Donkey). Brain responses that were modulated by word frequency of object names emerged in 160–240 and 260–340 msec, whereas lip–tongue contrasts were associated with 160–240 msec. Overall, recent results such as these highlight an early access to phonological codes in spoken production and hence constitute a challenge for most traditional models that have generally assumed that phonology is accessed later than semantics.

This Study

In the current study, rather than tracking relevant psycholinguistic variables such as semantic features or word frequency (as in Miozzo et al., 2015), we attempted to assess the relative time course of semantic and phonological access via the “blocked cyclic naming” paradigm, in which participants repeatedly name small sets of pictures in short experimental blocks, and the semantic or phonological context for a given block is manipulated. In its semantic form, this task is commonly used to explore lexical-semantic retrieval (Aristei, Melinger, & Abdel Rahman, 2011; Abdel Rahman & Melinger, 2007; Damian, Vigliocco, & Levelt, 2001; Kroll & Stewart, 1994). Pictures are arranged such that, within a block, all pictures stem from a single semantic category (“homogeneous”), or from various categories (“heterogeneous”). Across the study, each picture is named both in homogeneous and heterogeneous contexts and hence acts as its own control; the only aspect that varies is the homogeneous/heterogeneous semantic context. The typical finding is that objects are named slower in homogeneous than in heterogeneous blocks (Abdel Rahman & Melinger, 2009; Damian et al., 2001; Kroll & Stewart, 1994). This semantic blocking effect has been argued to originate at the lexical-semantic processing stage, although the mechanism underlying the semantic blocking effect is debated (e.g., Oppenheim, Dell, & Schwartz, 2010; Howard, Nickels, Coltheart, & Cole-Virtue, 2006; Damian & Als, 2005; Damian et al., 2001).

Rather than manipulating semantic context, one can also manipulate phonological context, such as whether or not responses within an experimental block share phonological properties. Here, the general finding is that word-initial overlap among items within a block facilitates responses (e.g., Chen & Chen, 2013; O'Seaghdha, Chen, & Chen, 2010; Roelofs, 2006; Damian & Bowers, 2003; Meyer, 1990, 1991), whereas noninitial overlap does not yield a benefit (e.g., Meyer, 1990, 1991). This phonological facilitation has conventionally been explained through partial planning of the response operating at the phonological processing level, which is possible in homogeneous but not in heterogeneous blocks (Roelofs, 1997). However, this explanation has recently been called into question by O'Seaghdha and Frazer (2014) who highlighted attentional influences on the effect, which raise the possibility that the effect resides outside the language system proper. In a recent article, Breining, Nozari, and Rapp (2016) reported highly intriguing results obtained with this task, which pose a further challenge with regard to how the effect informs us with regard to language production. As in numerous previous articles, they blocked response words within a set by form overlap. However, contrary to being limited to the word-initial position in previous studies, in the homogeneous blocks, responses were chosen such that overlap was distributed unpredictably across positions in words (e.g., cat-mat-cot-cap-map-mop). The critical finding was that this form of overlap slowed down naming latencies, compared to “heterogeneous” sets. This “similarity-based interference” effect is hence similar to the interference effect arising from semantic blocking summarized above, and the authors attributed both effects to a similar underlying principle of “incremental learning” (this notion will be unfolded in more detail in the Discussion).

In the current study, we manipulated semantic and phonological context within the same study and combined it with measurement of EEG. EPRs associated with the two types of context can provide important insights into the time course underlying word production. Our participants were native speakers of Chinese Mandarin. Semantic context was manipulated via blocking of semantic category membership. For the phonological manipulation, all picture names within a homogeneous block shared an atonal syllable but, for half of the items, the overlapping syllable was in the first position whereas, for the other half, it was in the second position. To the best of our knowledge, our study constitutes the first attempt to track the time course of phonological encoding with “inconsistent” overlap via EEG and to compare it directly with the semantic context effect. Behaviorally, both semantic and phonological contexts should result in interference relative to the unrelated baseline conditions. In EEG, we expected the semantic effect to begin at a relatively early time; as shown in Table 1, in similar studies, the semantic effect began around 200 msec post picture onset (the mean latency is 210 msec). The critical question was the onset of the phonological, relative to the semantic, effect. Models of spoken production, which assume some degree of “sequentiality” between semantic and phonological stages, would predict a later onset for phonological than for semantic effects. However, if both effects begin similarly early, that would constitute further difficulties for these accounts and favor a notion according to which access to semantic and to phonological codes in spoken production can occur in parallel.

Table 1. 
Summary of Studies Concerning Semantic Context Effect
StudyTechniqueTime Window (msec)Language
Anders et al. (2019intracerebral EEG 180–1000 French 
Aristei et al. (2011EEG 200–550 German 
Janssen et al. (2011EEG 220–450 Spanish 
Janssen et al. (2015EEG 250–400 and 500–750 Spanish 
Maess et al. (2002MEG 150–225 German 
Python et al. (2018EEG 270–315 French 
Wang et al. (2018EEG 200–550 Mandarin 
StudyTechniqueTime Window (msec)Language
Anders et al. (2019intracerebral EEG 180–1000 French 
Aristei et al. (2011EEG 200–550 German 
Janssen et al. (2011EEG 220–450 Spanish 
Janssen et al. (2015EEG 250–400 and 500–750 Spanish 
Maess et al. (2002MEG 150–225 German 
Python et al. (2018EEG 270–315 French 
Wang et al. (2018EEG 200–550 Mandarin 

METHODS

Participants

Following previous studies (e.g., Breining et al., 2016; Damian & Bowers, 2003), 24 native Mandarin Chinese speakers (17 women, mean age = 22 years) participated and were compensated for their time. All participants were right-handed, with normal or corrected-to-normal version and no history of language disorders. Participants gave informed consent, and the study was approved by the ethics committee of the Institute of Psychology, Chinese Academy of Sciences.

Materials and Design

For the semantic “type of overlap,” 16 objects were selected from four semantic categories and were arranged in a 4 × 4 matrix so that items in rows formed semantically homogeneous sets (i.e., items within a row stemmed from the same semantic category) whereas items in columns formed semantically heterogeneous sets (all items were from different categories). Items within a semantic homogeneous set were selected to minimize within-category visual similarity (e.g., 帆船, /fan1chuan2/, “steamboat”; 轿车, /jiao4che1/, “car”; 飞机, /fei1ji1/, “airplane”; 摩托, /mo2tuo2/, “motorbike”).

For the phonological “type of overlap,” another set of 16 objects were chosen such that, in phonologically homogeneous sets, all four items shared an atonal syllable but, for half items, the overlapping syllable was in word-initial position whereas, for the other half, it was in word-final position (e.g., 石头, /shi2tou0/, “rock”; 试管, /shi4guan3/, “tube”; 钥匙, /yao4shi0/, “key”; 电视, /dai4shi4/, “television”). For the corresponding phonologically heterogeneous sets, phonological overlap was avoided (e.g., 颜料, /yan2liao4/, “paint”; 帽子, /mao4zi0/, “hat”; 石头, /shi2tou0/, “rock”; 鲸鱼, /jing1yu2/, “whale”).

For the semantic combinations, phonological overlap was minimized; for the phonological combinations, semantic overlap was minimized. In all sets, orthographic overlap between items was avoided. Across the semantic and phonology conditions, pictures were statistically matched on various lexical properties.1 A full list of the materials is provided in Appendix as supplementary data.

Type of overlap (semantics vs. phonology) was manipulated as a within-participant and between-item variable, and context (homogeneous vs. heterogeneous) was varied as a within-participant and within-item variable. Half of the participants received the eight semantic blocks first, and the remaining half received the eight phonology blocks first. Furthermore, eight blocks (four homogeneous and four heterogeneous) were presented in an alternating sequence. The order of the four blocks was determined by a Latin Square design. Within each block, each item was presented for four cycles, resulting in 16 trials, in a pseudorandom order such that items were never repeated on adjacent trials. For each participant, the entire testing session included 256 trials (16 trials in each of 16 blocks).

Procedure

The experiment was run using the E-Prime 3.0 software (Psychology Software Tools), with a microphone connected to the computer recording vocal responses. Participants were tested individually and were instructed to name objects as fast and accurately as possible. Before each block, participants were first asked to familiarize themselves with the four pictures for that block, with the corresponding names printed underneath each object. Each trial started with a fixation (500 msec), and then a blank screen (500 msec) was followed by an object (2000 msec) in the center of the screen against a white background. The intertrial interval was 1000 msec. Participants received a practice block comprising four filler objects, followed by the 16 experimental blocks. The entire experiment lasted approximately 90 min per participant.

EEG Recordings and Preprocessing

EEG signals were recorded with 64 electrodes secured in an elastic cap (Electro Cap International) using Neuroscan 4.3. The vertical electrooculogram was monitored with electrodes placed above and below the left eye. The HEOG was recorded by a bipolar montage using two electrodes placed on the right and left external cantus. The left mastoid electrode served as a reference. All electrode impedances were below 5 kΩ. Electrophysiological signals were amplified with a band-pass filter of 0.05 and 70 Hz (sampling rate = 500 Hz).

The EEGLAB toolbox based on MATLAB was used for the following procedure of preprocessing EEG signals: off-line filter with a high-pass cutoff of 0.1 Hz and a low-pass cutoff of 30 Hz; removal of ocular, muscle, motor artifacts, and linear noise using independent component analyses (Jung et al., 2000) on the segmented data (−0.8 to 1.5 sec relative to the picture onset); manual rejection of epochs with extensive fluctuation and signals below/above ± 70 μV; and off-line rereferencing against the average reference. The EEG was segmented into 600-msec epochs relative to picture onset that included a 100-msec prestimulus baseline and a 500-msec poststimulus interval.

Response Latency Analysis

Naming latencies were analyzed using a linear mixed-effects model (Baayen, Davidson, & Bates, 2008) with the package lme4 (Bates & Maechler, 2009). We constructed a model including the main effects of Type of overlap, Context, and Cycle, as well as their interactions (model. matrix = ∼type × context × cycle). Furthermore, to evaluate the context effects along with cycles within each type, another model was constructed by including the main effect of Type, and simple effects of Context, Cycle, and interactions between Context and Cycle in each type (model. matrix = ∼type/(context × cycle)). In each model, a full random structure was implemented, with random intercepts and random slopes over participants and items. When the model failed to converge, a reduction procedure with the principle component analysis was conducted (Bates, Kliegl, Vasishth, & Baayen, 2015) with the rePCA function in the package RePsychLing (Baayen, Bates, Kliegl, & Vasishth, 2015) until the model could be supported by the data. We report p values derived from the t values using the normal approximation (Mirman, 2014).

ERP Analysis

Two types of analyses were conducted on the ERP data. First, onset latency analyses were performed. For each type of overlap (semantic or phonological), ERPs for homogeneous and heterogeneous conditions were compared by running t tests on all electrodes at every sampling point (every 2 msec) from −100 to 500 msec relative to picture presentation, with the aim of identifying the latency at which the ERPs started to diverge significantly from each other. To protect against problems associated with multiple comparisons, we performed onset latency analyses using a method developed by Guthrie and Buchwald (1991; see Qu & Damian, 2020; Qu, Zhang, & Damian, 2016; Strijkers et al., 2010; Costa, Strijkers, Martin, & Thierry, 2009; Thierry, Cardebat, & Demonet, 2003, for use of this method in recent studies). This method estimates how long an interval of consecutive significant points can be expected by chance via computer simulations based on autocorrection coefficients, sample sizes, and sampling interval length. If the observed number of consecutive significant time intervals is longer than the significant interval expected by chance, this indicate a statistically significant interval, and the onset point of the consecutive significant points is taken as the onset of the corresponding effect. In this analysis, we averaged the onset latencies of those electrodes, which showed significant intervals in each type, and regarded the averaged values as the onsets of the two effects.

We examined whether the onset latencies of semantics versus phonology were significantly different from each other using the jackknife approach (Ulrich & Miller, 2001; Miller, Patterson, & Ulrich, 1998). A jackknife waveform was computed for each participant i (i = 1 … n, where n is the number of participants) by temporarily omitting participant i and computing the grand average of the difference in ERPs from the remaining n − 1 participants. The jackknife onset latency was determined by the t tests at every sampling point on the electrodes, which showed significant intervals. To determine the statistical difference of the latencies of semantic effect and phonological effect, the onset latencies of the two effects were compared with a repeated-measures ANOVA (Type of overlap as a within-participant factor). The same procedure was conducted to compare the time points where the two effects reached their maximum value. A corrected F value (Fc = F/(n − 1)2) was used to determine statistical significance.

Second, mean amplitude analysis was performed. Five time windows for mean amplitude analysis were selected on the basis of visible peaks in the grand average ERP waveforms, combined with the consideration of the selection of time windows in previous production studies (e.g., Strijkers et al., 2010; Costa et al., 2009): [0–180 msec], [180–250 msec] (P2), [280–320 msec] (N2), [320–380 msec] (P3), and [380-500 msec]. For each type in each time window, mean amplitudes of the selected electrodes were entered into repeated-measures ANOVA. False discovery rate correction (Genovese & Wasserman, 2002; Benjamini & Hochberg, 1995) was applied on the obtained p values to control the potential Type I error.

RESULTS

In blocked cyclic naming tasks, performance on the first cycle within an experimental block often differs from the remaining cycles. The effect of semantic blocking, which is overall interfering, is often facilitatory on the first cycle (e.g., Navarrete, Del Prato, Peressotti, & Mahon, 2014; Schnur, Schwartz, Brecher, & Hodgson, 2006; Belke, Meyer, & Damian, 2005), and occasionally absent (e.g., Damian & Als, 2005). This pattern has generated an extensive discussion of the possibility that semantic effects on Cycle 1 and on subsequent cycles might be caused by different underlying psychological and neural mechanisms (e.g., Navarrete, Del Prato, & Mahon, 2012; Belke, 2008). Recently, Python, Fargier, and Laganaro (2018) combined semantic blocking with EEG and attributed the effect on the first cycle to postlexical interactive phonological-semantic processes and/or to self monitoring, whereas the effect on subsequent cycles reflected lexical competition. Effects of phonological overlap in blocked cyclic naming have been less well investigated, but, in their pioneering study, Breining et al. (2016) also suggested that effects on the first cycle might differ from the one on the remaining cycles within a block. For this reason and in accordance with previous studies such as Breining et al., we considered only the results from Cycles 2 to 4 for the analyses of RTs and ERPs. However, Figure 1 below shows performance results from all cycles, and analyses of behavioral and ERP results were overall similar with and without Cycle 1.

Figure 1. 

Mean response times by type of overlap (semantics vs. phonology) and context (homogeneous vs. heterogeneous). Error bars indicate SEM.

Figure 1. 

Mean response times by type of overlap (semantics vs. phonology) and context (homogeneous vs. heterogeneous). Error bars indicate SEM.

Data with missing recordings (1.2%), incorrect responses (0.6%), and latencies shorter than 300 msec or longer than 1500 msec (2.9%), and beyond 2.5 SDs (1.1%) were excluded from the behavioral and ERP analyses. For the ERP analyses, a further 8.7% of trials were excluded because of artifacts. In total, ERP analyses were based on an average of 44 segments per condition (semantic homogeneous, 43; semantic heterogeneous, 44; phonological homogeneous, 44; phonological heterogeneous, 45).

Behavioral Results

As shown in Figure 1, for both types of overlap, responses in the homogeneous context were longer than that in the heterogeneous context (semantics: 562 vs. 525 msec; phonology: 531 vs. 517 msec). The linear mixed-effects model analysis showed a significant main effect of Context (β = 26.18, SE = 5.49, t = 4.77, p < .001) and a marginally significant interaction between type and context (β = 22.11, SE = 11.57, t = 1.91, p = .056), suggesting a somewhat larger context effect in the semantic than in the phonological type of overlap. The main effect of Type of overlap was not significant (semantics: 543 msec, phonology: 524 msec, β = 18.70, SE = 11.02, t = 1.7, p = .09). Further analyses for each type of overlap revealed a context effect for both the semantics and phonology conditions (semantics: 562 vs. 525 msec, β = 37.11, SE = 7.56, t = 4.91, p < .001; phonology: 531 vs. 517 msec, β = 15.02, SE = 4.17, t = 3.61, p < .001). Only the semantic effect interacted with Cycles (β = 20.57, SE = 10.12, t = 2.03, p = .042), whereas the phonological effect did not (p > .4). No analysis was performed on accuracy as the error rate was extremely low (∼0.6%).

ERP Results

Grand average ERP waveforms are displayed in Figure 2 for the semantic and phonological conditions averaged from the data of all participants on selected electrodes. Target pictures elicited an expected P1/N1/P2 ERP component in all conditions, and the homogeneous context in both conditions produced less positive amplitudes than the heterogeneous context did.

Figure 2. 

(A) Grand average ERPs for homogeneous (solid lines) and heterogeneous (dash lines) context in the semantic type of overlap. (B) Grand average ERPs for homogeneous (solid lines) and heterogeneous (dash lines) context in the phonological type of overlap. The onset of a target object is represented by 0 msec. Semantic and phonological overlap modulated ERPs in the time window of 180–380 msec.

Figure 2. 

(A) Grand average ERPs for homogeneous (solid lines) and heterogeneous (dash lines) context in the semantic type of overlap. (B) Grand average ERPs for homogeneous (solid lines) and heterogeneous (dash lines) context in the phonological type of overlap. The onset of a target object is represented by 0 msec. Semantic and phonological overlap modulated ERPs in the time window of 180–380 msec.

Onset Latency Analysis

Separately for the semantics and the phonology conditions, ERPs corresponding to the homogeneous and heterogeneous conditions were compared by running t tests at every 2 msec starting −100 to 500 msec relative to picture onset over all 62 electrodes. Onset latencies were computed on averages of those electrodes in which the observed number of consecutive significant time points was larger than the critical run length to determine statistical significance. For the semantic effect, the averaged splitting point computed from individual splitting point estimates (nine electrodes: CP1, CP3, P1, P2, P3, P4, PO3, PO4, and PZ) was 175 msec after picture onset. The averaged splitting point for the phonological effect (six electrodes: P1, P2, PZ, POZ, OZ, and PO4) was 192 msec after picture onset. Results of the jackknife method showed that the time points at which the effects started to emerge was not significantly different between the semantic and the phonological conditions (Fcorrected < 1). Neither did the time points differ significantly at which the semantic and phonological effects reached their respective maximum value (semantics: 243 msec, phonology: 272 msec, Fcorrected < 1).

Mean Amplitude Analyses

Six ROIs were selected to probe the scalp distribution of ERP differences: left-anterior (F3, FC3, FC5), middle-anterior (FZ, FCZ, CZ), right-anterior (F4, FC4, FC6), left-posterior (CP3, P3, P5), middle-posterior (CPZ, PZ, POZ), and right-posterior (CP4, P4, P6). For each condition, the main results of the omnibus ANOVA, conducted separately for each of the five time intervals (0–180, 180–250, 280–320, 320–380, and 380–500 msec), were as follows. Moreover, critical time windows were aggregated into a larger time range, and additional analyses were conducted on the larger time range, that is, 180–380 msec.

In the early time window from 0 to 180 msec after picture onset, for the semantics condition, only the interaction between Context and Laterality was significant, F(2, 46) = 3.82, p = .03. However, post hoc analysis in each ROI did not show any significant Context effects (all ps > .12). For the phonology condition, neither the main effect of Context nor interactions involving Context were significant (ps > .13), and none of the ROIs showed Context effect (ps > .35). In the P2 range (180–250 msec), the interaction between Context, Anteriority, and Laterality was significant in the semantics condition, F(2, 46) = 5.39, p = .007, and marginally significant in the phonology condition, F(2, 46) = 2.75, p = .077. Critically, post hoc analysis revealed significant semantic effects in the left posterior, t(23) = −3.59, p = .002; middle posterior, t(23) = −3.69, p = .002; and right posterior, t(23) = −2.49, p = .02, and a significant phonological effect in the middle posterior region, t(23) = −3.26, p = .01. In the following N2 range (280–320 msec), the interaction between Context, Anteriority, and Laterality was significant in the semantics Condition, F(2, 46) = 3.51, p = .040, and the phonology condition, F(2, 46) = 3.82, p = .03. The Context effects were marginally significant in the middle posterior, t(23) = −2.44, p = .069, and right posterior, t(23) = −1.97, p = .091, in the semantics condition, and middle posterior in the phonology condition, t(23) = −2.52, p = .058. In the P3 range (320–380 msec), the interaction between Context, Anteriority, and Laterality was marginally significant in the semantics condition, F(2, 46) = 2.76, p = .075, and significant in the phonology condition, F(2, 46) = 3.39, p = .044. Left posterior region in the semantics condition, t(23) = −2.92, p = .023, and middle posterior region in the phonology condition, t(23) = −2.77, p = .032, demonstrated significant Context effects. In the late time window (380–500 msec), neither the main effect of context, nor interactions involved context were significant (ps > 0.1), and none of ROIs showed context effect (ps > 0.35). Hence, both semantic and phonological relatedness showed modulation of ERPs in the same time windows, starting about 180 msec and ending at about 380 msec after picture onset.

This pattern of ERP effects was confirmed via additional analyses on a larger time window, that is, 180–380 msec. In the semantics condition, the interaction among Anteriority, Laterality, and Context was significant, F(2, 46) = 5.56, p = .007. Further analysis at each ROI (p values false discovery rate corrected) revealed a significant semantic effect in the left posterior, t(23) = −3.15, p = .013; middle posterior, t(23) = −2.83, p = .014; and right posterior regions, t(23) = −2.07, p = .049. Similarly, the phonology condition also demonstrated a three-way interaction among Anteriority, Laterality, and Context, F(2, 46) = 3.56, p = .036. The phonology effect was significant in the middle posterior region, t(23) = −3.06, p = .016. This pattern of ERP effects were further confirmed via additional analyses on a larger set of 18 electrodes of the central-posterior region (CZ, CPZ, PZ, POZ, C1, CP1, P1, C2, CP2, P2, C3, CP3, P3, PO3, C4, CP4, P4, and PO4; semantics: t(23) = −3.68, p < .001; phonology: t(23) = −2.36, p = .013). Hence, ERP data showed that semantic overlap and phonological overlap modulated ERPs in a time window of 180–380 msec, across the central-posterior regions.

DISCUSSION

With the EEG technique, we investigated the time course of semantic and phonological processes in spoken word production via the blocked cyclic naming paradigm. Objects were embedded within lists of same-category items, or lists of phonologically overlapping items (or within lists of unrelated items). Consistent with previous studies (Belke et al., 2005; Damian et al., 2001), objects were named more slowly in the context of same-category items than when stemming from various semantic categories. Phonological overlap in our study could occur either in first or second word position of responses within a homogeneous block, and this also resulted in slower naming responses than in the corresponding unrelated baseline condition. This latter finding—phonological interference—poses an interesting contrast with numerous previous studies in which word-initial phonological overlap was shown to result in facilitation (e.g., Meyer, 1990), and it replicates the finding of Breining et al. (2016) with English materials, as well as our own (Qu, Feng, & Damian, under review) with Chinese stimuli.

Most importantly for present purposes, the ERP data showed that semantic and phonological overlap modulated ERPs in similar time windows (180–380 msec). Precise temporal analysis revealed that the semantic effect emerged with an onset of 175 msec and the phonological effect had an onset of 192 msec. The two effects did not differ significantly regarding the time points where the effects began, nor where they reached maximum. Semantic relatedness evoked less positive ERPs compared to the baseline condition, which is consistent with previous studies (Anders et al., 2019; Python et al., 2018; Wang, Shao, Chen, & Schiller, 2018; Janssen, Hernández-Cabrera, van der Meij, & Barber, 2015; Maess, Friederici, Damian, Meyer, & Levelt, 2002; but see Janssen, Carreiras, & Barber, 2011). Similarly, ERPs were also less positive in the phonologically related context than in the baseline condition.

The most striking aspect of our EEG results is that, onset latencies, as well as the time interval, for the two types of context effects were almost identical. Evidently, semantic processing takes place in parallel with phonological encoding, within approximately 180 msec after picture onset. This finding appears to contradict models of word production, which conceive of semantic and phonological encoding as discrete processing stages (Levelt et al., 1999; the seriality assumption is also critical to Indefrey's, 2002, and Indefrey & Levelt's, 2004, time estimates of processing stages in spoken production). According to these models, lexical-semantic activation precedes phonological encoding, so in an EEG study, the time window that indexes lexical-semantics should precede the one associated with access to phonology. There are several potential explanations for the empirical pattern. One possibility is that activation strongly cascades throughout the network; hence, the phonological encoding system receives activation almost as soon as processing begins at the lexical-semantic stage. Therefore, our finding that semantic and phonological effects emerged at the similar time interval with slightly earlier semantic effect might be compatible with models of lexical access, which incorporate cascading activation (and potentially, feedback) between lexical-semantic and phonological encoding (Rapp & Goldrick, 2000; Peterson & Savoy, 1998; Dell & O'Seaghdha, 1992; Dell, 1986).

An alternative and new perspective is offered by neural assembly models where all components of speech production are organized in a single functional unit in the way of connected neural populations (Strijkers, 2016a, 2016b; Strijkers & Costa, 2016; Pulvermüller, 1999). These components could be activated synchronously or near-simultaneously. A critical difference between the cascading and neural assembly models concerns whether the interaction is restricted to adjacent levels. The cascading models predict that the cascading activation only emerges between adjacent stages (e.g., from visual processing to semantics; from semantics to phonology), whereas the nonhierarchical neural assembly models predict that interactivity likely reaches beyond adjacent layers (e.g., from visual processing to phonology). In this study, the semantic effect and phonological effects are associated with two neighboring processing levels (lexical-semantic and phonological encoding), and thus, our results are not able to distinguish between the two types of models. However, it should be noted that cascaded interactive processing models generally assume that adjacent processing stages are separated by functional delays of roughly 100 msec (Dell & O'Seaghdha, 1992). The tiny ∼20 msec of onset latency difference between semantics and phonology is not quantitatively compatible with this assumption. Therefore, the current findings are not easily explained by cascading models and instead favor neural assembly models.

The time window that hosts semantic effects in our study (180–380 msec) is compatible with findings from ERP studies with the blocked cyclic naming paradigm (Python et al., 2018; Maess et al., 2002; Aristei et al., 2011). In addition, this time course is also consistent with studies using other overt picture naming tasks. In a picture naming study, Strijkers et al. (2010) observed a “word frequency effect” (an effect that is assumed to arise from lexical-semantic retrieval) that started at 180 msec after picture presentation. Miozzo et al. (2015) used a multiple linear regression approach to MEG analysis and found early effects (around 150 msec) of variables related to semantic processing. It has been shown that the time course of semantic processing is independent of participants' response speed (Laganaro, Valente, & Perret, 2012) and naming modalities, that is, speaking versus writing (Perret & Laganaro, 2012). The onset of the semantic effect found in this study (175 msec) is broadly in agreement with the estimated onset time of lexical-semantic encoding proposed by Indefrey and Levelt (2004; 175 msec) and the update provided by Indefrey (2011; 200 msec).

By contrast, the onset of the phonological effect in our study (192 msec) is earlier than predicted from the estimates by Indefrey and Levelt (250 msec) and Indefrey (275 msec). However, such an early onset is broadly in line with recent findings from related tasks, which have suggested that phonology is accessed around 200 msec (Qu & Damian, 2020; Zhang, Yu, Zhang, Jin, & Li, 2018; Strijkers et al., 2017; Wang, Wong, Wang, & Chen, 2017; Miozzo et al., 2015; Yu et al., 2014; Qu et al., 2012). For example, overlap of a single phoneme between successive words modulated ERPs in a time window of 200–300 msec in a colored picture naming task (Qu et al., 2012) and 180–300 msec in a picture–picture priming task (Yu et al., 2014). However, it is worth noting that EEG results from the PWI task have suggested a later time course of phonological encoding. For instance, Zhu et al. (2015) and Wong, Wang, Ng, and Chen (2016) found phonologically based ERP effects at 450–600 msec and 500–600 msec, respectively; Dell'Acqua et al. (2010) observed form-based ERP effects in a slightly earlier time interval of 250–400 msec after picture onset, but this interval is still later than that found in this study. One possible reason for this discrepancy might be that, in picture–word experiments, processing of a distractor word itself delays the normally rapid phonological encoding of the spoken utterance.

How does the semantic context effect relate to spoken word production? Interfering effects of semantic context in the blocked cyclic naming paradigm were initially interpreted via competitive selection among a set of words that was co-activated by the spreading activation mechanism (Damian et al., 2001). However, a subsequent study (Damian & Als, 2005) showed that the effect did not diminish when several unrelated items intervened between critical naming responses. A semantic context effect that persists over a relatively long period is incompatible with an explanation of the effect based on spreading activation, which is generally assumed to be short-lived. An alternative account (Oppenheim et al., 2010; Howard et al., 2006; Damian & Als, 2005) is that engagement of the semantic-lexical pathway leads to “incremental learning,” that is, a slight but persistent increment in the connection weights between the representational layers. This will increase efficiency when the same item is named again (“repetition priming”; e.g., Cave, 1997), but, at the same time, it will slightly slow down the naming of semantically related items. Howard et al. (2006) presented a computational model in which semantic features were connected to nodes within a lexical layer, and the latter inhibited each other. Slight increments in the connections simulated the cumulative semantic inhibition effect. Oppenheim et al. (2010) provided an extensive computational simulation of these phenomena. They introduced a model in which semantic features were connected to lexical nodes and semantic-to-lexical links were modified on each trial such that weights from semantically active features to target lexical nodes were increased, whereas weights to all other lexical nodes were decreased. This principle of “competitive learning” was able to generate both cumulative semantic inhibition and cyclic semantic blocking, but intriguingly, lexical competition was not necessary to do so (however, see Roelofs, 2018, for a more recent computational simulation of various semantic effects in spoken word production that retains the notion of competitive lexical selection as a central architectural feature).

This study found that, just like semantic overlap, phonological form overlap across different word positions caused interference, hence replicating a pattern first demonstrated by Breining et al. (2016). In a recent study of ours, we (Qu et al., under review) have demonstrated that phonological interference was found even when (semantically and phonologically) unrelated filler pictures were interleaved with the critical targets. Hence, the phonological interference effect is reasonably long-lasting. For this reason, we interpret this finding as implying that the phonological effect is also because of competitive incremental learning (as was claimed for the semantic effect; see above), but incremental learning occurs between lexical and phonological representations for phonology. According to this mechanism, object naming elicits a slight and persistent modification of connections between lexical and phonological representations. For example, consider a mini-scenario with only two phonologically related objects (“cat”-“cap”). On trial N − 1, the lexical code for “cat” and its corresponding phonological codes are activated. Via feedback from the phonological to the lexical level, phonologically related lexical codes (e.g., “cap”) will be co-activated. Following the trial N − 1, all links between the target code and the phonological level are strengthened, and connections between the lexical node “cap” and the phonological level are decremented. On a subsequent trial, if the phonological related item “cap” is to be named, some of the connections between the lexical node and the phonological codes have been already weakened on the previous trial, and so naming times and/or accuracy for a naming response to “cap” will be detrimentally affected. Speculatively, incremental learning constitutes a universal principle in semantic-to-lexical and lexical-to-phonology mappings.

As far as ERP components are concerned, as reported previously for word frequency, cognate status (Strijkers et al., 2010; see also Strijkers, Baus, Runnqvist, FitzPatrick, & Costa, 2013), and cumulative semantic interference in picture naming (Costa et al., 2009), we found that both the semantic effect and the phonological effect elicited the same modulations of electrophysiological responses at P2, N2, and P3. Perhaps the most interesting ERP component is P2, which appears to be modulated by various lexical variables such as lexical frequency, cognate status, and semantic interference and is thus assumed to reflect lexical access. Similarly, our study reported the P2 modulations induced by both semantic and phonological contexts. Based on these findings, we emphasize the effectiveness of the P2 component as an index of lexical access in speech production.

In summary, our behavioral results showed that both semantic and phonological overlap between spoken response words within an experimental block of picture naming responses produced longer naming latencies, compared to the baseline conditions. Our ERP results provide evidence for the claim that lexical-semantic and phonological processing proceed largely in parallel. These findings will provide important constraints on models of spoken word production.

APPENDIX: MATERIALS USED IN THE EXPERIMENT

  Heterogeneous Lists
Semantics Homogeneous Lists 摩托 (mo2tuo2) (motorbike) 飞机 (fei1ji1) (airplane) 轮船 (lun2chuan2) (steamboat) 轿车 (jiao4che1) (car) 
鼻子 (bi2zi0) (nose) 眼睛 (yan3jing1) (eye) 手指 (shou3zhi3) (finger) 耳朵 (er3duo0) (ear) 
樱桃 (ying1tao2) (cherry) 菠萝 (bo1luo2) (pineapple) 西瓜 (xi1gua1) (watermelon) 柿子 (shi4zi0) (persimmon) 
围巾 (wei2jin1) (scarf) 裤子 (ku4zi0) (pants) 衬衫 (chen4shan1) (shirt) 背心 (bei4xin1) (vest) 
  
Phonology Homogeneous Lists 圣经 (sheng4jing1) (bible) 镜子 (jing4zi0) (mirror) 鲸鱼 (jing1yu2) (whale) 水井 (shui3jing3) (water well) 
电视 (dian4shi4) (television) 试管 (shi4guan3) (test tube) 石头 (shi2tou2) (stone) 钥匙 (yao4shi0) (key) 
屋檐 (wu1yan2) (eave) 燕子 (yan4zi0) (swallow) 颜料 (yan2liao4) (paint) 香烟 (xiang1yan1) (cigarette) 
熊猫 (xiong2mao1) (panda) 毛巾 (mao2jin1) (towel) 帽子 (mao4zi0) (cap) 长矛 (chang2mao2) (spear) 
  Heterogeneous Lists
Semantics Homogeneous Lists 摩托 (mo2tuo2) (motorbike) 飞机 (fei1ji1) (airplane) 轮船 (lun2chuan2) (steamboat) 轿车 (jiao4che1) (car) 
鼻子 (bi2zi0) (nose) 眼睛 (yan3jing1) (eye) 手指 (shou3zhi3) (finger) 耳朵 (er3duo0) (ear) 
樱桃 (ying1tao2) (cherry) 菠萝 (bo1luo2) (pineapple) 西瓜 (xi1gua1) (watermelon) 柿子 (shi4zi0) (persimmon) 
围巾 (wei2jin1) (scarf) 裤子 (ku4zi0) (pants) 衬衫 (chen4shan1) (shirt) 背心 (bei4xin1) (vest) 
  
Phonology Homogeneous Lists 圣经 (sheng4jing1) (bible) 镜子 (jing4zi0) (mirror) 鲸鱼 (jing1yu2) (whale) 水井 (shui3jing3) (water well) 
电视 (dian4shi4) (television) 试管 (shi4guan3) (test tube) 石头 (shi2tou2) (stone) 钥匙 (yao4shi0) (key) 
屋檐 (wu1yan2) (eave) 燕子 (yan4zi0) (swallow) 颜料 (yan2liao4) (paint) 香烟 (xiang1yan1) (cigarette) 
熊猫 (xiong2mao1) (panda) 毛巾 (mao2jin1) (towel) 帽子 (mao4zi0) (cap) 长矛 (chang2mao2) (spear) 

The number denotes the tone for the preceding syllable. There are four tones in Mandarin. The number 0 represents a neutral tone. The overlapped syllable is underlined.

Reprint requests should be sent to Qingqing Qu, Key Laboratory of Behavioral Science, Institute of Psychology, Chinese Academy of Sciences, 16 Lincui Road, Chaoyang District, Beijing, 100101, China, or via e-mail: quqq@psych.ac.cn.

Funding Information

Qingqing Qu, Youth Innovation Promotion Association of the Chinese Academy of Sciences (http://dx.doi.org/10.13039/501100004739), Youth Talent Project of the China Association for Science and Technology, National Natural Science Foundation of China (http://dx.doi.org/10.13039/501100001809), grant numbers: 31771212, 62061136001.

Diversity in Citation Practices

A retrospective analysis of the citations in every article published in this journal from 2010 to 2020 has revealed a persistent pattern of gender imbalance: Although the proportions of authorship teams (categorized by estimated gender identification of first author/last author) publishing in the Journal of Cognitive Neuroscience (JoCN) during this period were M(an)/M = .408, W(oman)/M = .335, M/W = .108, and W/W = .149, the comparable proportions for the articles that these authorship teams cited were M/M = .579, W/M = .243, M/W = .102, and W/W = .076 (Fulvio et al., JoCN, 33:1, pp. 3–7). Consequently, JoCN encourages all authors to consider gender balance explicitly when selecting which articles to cite and gives them the opportunity to report their article's gender citation balance.

Note

1. 

Stimuli in the semantic and phonological conditions were matched on the following variables: word frequency (p = .081), visual complexity (p = .961), naming agreement (p = .602), and stroke number (p = .322). Values were taken from the work of Liu, Hao, Li, and Shu (2011).

REFERENCES

Abdel Rahman
,
R.
, &
Melinger
,
A.
(
2007
).
When bees hamper the production of honey: Lexical interference from associates in speech production
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
33
,
604
614
.
Abdel Rahman
,
R.
, &
Melinger
,
A.
(
2009
).
Semantic context effects in language production: A swinging lexical network proposal and a review
.
Language and Cognitive Processes
,
24
,
713
734
.
Abdel Rahman
,
R.
, &
Sommer
,
W.
(
2003
).
Does phonological encoding in speech production always follow the retrieval of semantic knowledge? Electrophysiological evidence for parallel processing
.
Cognitive Brain Research
,
16
,
372
382
.
Anders
,
R.
,
Llorens
,
A.
,
Dubarry
,
A.-S.
,
Trébuchon
,
A.
,
Liégeois-Chauvel
,
C.
, &
Alario
,
F.-X.
(
2019
).
Cortical dynamics of semantic priming and interference during word production: An intracerebral study
.
Journal of Cognitive Neuroscience
,
31
,
978
1001
.
Aristei
,
S.
,
Melinger
,
A.
, &
Abdel Rahman
,
R.
(
2011
).
Electrophysiological chronometry of semantic context effects in language production
.
Journal of Cognitive Neuroscience
,
23
,
1567
1586
.
Baayen
,
R. H.
,
Bates
,
D.
,
Kliegl
,
R.
, &
Vasishth
,
S.
(
2015
).
RePsychLing: Data sets from psychology and linguistics experiments (R package version 0.0.4)
. .
Baayen
,
R. H.
,
Davidson
,
D. J.
, &
Bates
,
D. M.
(
2008
).
Mixed-effects modeling with crossed random effects for subjects and items
.
Journal of Memory and Language
,
59
,
390
412
.
Bates
,
D.
,
Kliegl
,
R.
,
Vasishth
,
S.
, &
Baayen
,
H.
(
2015
).
Parsimonious mixed models
.
arXiv preprint arXiv:1506.04967
.
Bates
,
D.
, &
Maechler
,
M.
(
2009
).
lme4: Linear mixed-effects models using S4 classes. R package version 0.999375-31
. .
Belke
,
E.
(
2008
).
Effects of working memory load on lexical-semantic encoding in language production
.
Psychonomic Bulletin & Review
,
15
,
357
363
.
Belke
,
E.
,
Meyer
,
A. S.
, &
Damian
,
M. F.
(
2005
).
Refractory effects in picture naming as assessed in a semantic blocking paradigm
.
Quarterly Journal of Experimental Psychology
,
58
,
667
692
.
Benjamini
,
Y.
, &
Hochberg
,
Y.
(
1995
).
Controlling the false discovery rate: A practical and powerful approach to multiple testing
.
Journal of the Royal Statistical Society, Series B, Methodological
,
57
,
289
300
.
Breining
,
B.
,
Nozari
,
N.
, &
Rapp
,
B.
(
2016
).
Does segmental overlap help or hurt? Evidence from blocked cyclic naming in spoken and written production
.
Psychonomic Bulletin & Review
,
23
,
500
506
.
Camen
,
C.
,
Morand
,
S.
, &
Laganaro
,
M.
(
2010
).
Re-evaluating the time course of gender and phonological encoding during silent monitoring tasks estimated by ERP: Serial or parallel processing?
Journal of Psycholinguistic Research
,
39
,
35
49
.
Cave
,
C. B.
(
1997
).
Very long-lasting priming in picture naming
.
Psychological Science
,
8
,
322
325
.
Chen
,
T.-M.
, &
Chen
,
J.-Y.
(
2013
).
The syllable as the proximate unit in Mandarin Chinese word production: An intrinsic or accidental property of the production system?
Psychonomic Bulletin & Review
,
20
,
154
162
.
Costa
,
A.
,
Strijkers
,
K.
,
Martin
,
C.
, &
Thierry
,
G.
(
2009
).
The time course of word retrieval revealed by event-related brain potentials during overt speech
.
Proceedings of the National Academy of Sciences, U.S.A.
,
106
,
21442
21446
.
Damian
,
M. F.
, &
Als
,
L. C.
(
2005
).
Long-lasting semantic context effects in the spoken production of object names
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
31
,
1372
1384
.
Damian
,
M. F.
&
Bowers
,
J. S.
(
2003
).
Effects of orthography on speech production in a form-preparation paradigm
.
Journal of Memory and Language
,
49
,
119
132
.
Damian
,
M. F.
, &
Martin
,
R. C.
(
1999
).
Semantic and phonological codes interact in single word production
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
25
,
345
361
.
Damian
,
M. F.
,
Vigliocco
,
G.
, &
Levelt
,
W. J.
(
2001
).
Effects of semantic context in the naming of pictures and words
.
Cognition
,
81
,
B77
B86
.
Dell
,
G. S.
(
1986
).
A spreading-activation theory of retrieval in sentence production
.
Psychological Review
,
93
,
283
321
.
Dell
,
G. S.
, &
O'Seaghdha
,
P. G.
(
1992
).
Stages of lexical access in language production
.
Cognition
,
42
,
287
314
.
Dell
,
G. S.
,
Schwartz
,
M. F.
,
Martin
,
N.
,
Saffran
,
E. M.
, &
Gagnon
,
D. A.
(
1997
).
Lexical access in aphasic and nonaphasic speakers
.
Psychological Review
,
104
,
801
838
.
Dell'Acqua
,
R.
,
Sessa
,
P.
,
Peressotti
,
F.
,
Mulatti
,
C.
,
Navarrete
,
E.
, &
Grainger
,
J.
(
2010
).
ERP evidence for ultra-fast semantic processing in the picture–word interference paradigm
.
Frontiers in Psychology
,
1
,
177
.
Garrett
,
M. F.
(
1975
).
The analysis of sentence production
. In
G.
Bower
(Ed.),
The psychology of learning and motivation
(Vol. 9, pp.
133
177
).
New York
:
Academic Press
.
Genovese
,
C.
, &
Wasserman
,
L.
(
2002
).
Operating characteristics and extensions of the false discovery rate procedure
.
Journal of the Royal Statistical Society, Series B, Statistical Methodology
,
64
,
499
517
.
Glaser
,
M. O.
, &
Glaser
,
W. R.
(
1982
).
Time course analysis of the Stroop phenomenon
.
Journal of Experimental Psychology: Human Perception and Performance
,
8
,
875
894
.
Glaser
,
W. R.
, &
Düngelhoff
,
F. J.
(
1984
).
The time course of picture–word interference
.
Journal of Experimental Psychology: Human Perception and Performance
,
10
,
640
654
.
Guthrie
,
D.
, &
Buchwald
,
J. S.
(
1991
).
Significance testing of difference potentials
.
Psychophysiology
,
28
,
240
244
.
Howard
,
D.
,
Nickels
,
L.
,
Coltheart
,
M.
&
Cole-Virtue
,
J.
(
2006
).
Cumulative semantic inhibition in picture naming: Experimental and computational studies
.
Cognition
,
100
,
464
482
.
Indefrey
,
P.
(
2002
).
Listen und Regeln: Erwerb und Repräsentation der schwachen Substantivdeklination des Deutschen
(doctoral dissertation)
.
Heinrich-Heine-Universität Düsseldorf
.
Indefrey
,
P.
(
2011
).
The spatial and temporal signatures of word production components: A critical update
.
Frontiers in Psychology
,
2
,
255
.
Indefrey
,
P.
, &
Levelt
,
W. J. M.
(
2004
).
The spatial and temporal signatures of word production components
.
Cognition
,
92
,
101
144
.
Janssen
,
N.
,
Carreiras
,
M.
, &
Barber
,
H. A.
(
2011
).
Electrophysiological effects of semantic context in picture and word naming
.
Neuroimage
,
57
,
1243
1250
.
Janssen
,
N.
,
Hernández-Cabrera
,
J. A.
,
van der Meij
,
M.
, &
Barber
,
H. A.
(
2015
).
Tracking the time course of competition during word production: Evidence for a post-retrieval mechanism of conflict resolution
.
Cerebral Cortex
,
25
,
2960
2969
.
Jung
,
T. P.
,
Makeig
,
S.
,
Humphries
,
C.
,
Lee
,
T. W.
,
Mckeown
,
M. J.
,
Iragui
,
V.
, et al
(
2000
).
Removing electroencephalographic artifacts by blind source separation
.
Psychophysiology
,
37
,
163
178
.
Kroll
,
J. F.
, &
Stewart
,
E.
(
1994
).
Category interference in translation and picture naming: Evidence for asymmetric connections between bilingual memory representations
.
Journal of Memory and Language
,
33
,
149
174
.
Laganaro
,
M.
, &
Perret
,
C.
(
2011
).
Comparing electrophysiological correlates of word production in immediate and delayed naming through the analysis of word age of acquisition effects
.
Brain Topography
,
24
,
19
29
.
Laganaro
,
M.
,
Valente
,
A.
, &
Perret
,
C.
(
2012
).
Time course of word production in fast and slow speakers: A high density ERP topographic study
.
Neuroimage
,
59
,
3881
3888
.
Levelt
,
W. J.
,
Roelofs
,
A.
, &
Meyer
,
A. S.
(
1999
).
A theory of lexical access in speech production
.
Behavioral and Brain Sciences
,
22
,
1
38
.
Liu
,
Y.
,
Hao
,
M.
,
Li
,
P.
, &
Shu
,
H.
(
2011
).
Timed picture naming norms for Mandarin Chinese
.
PLoS One
,
6
,
e16505
.
Maess
,
B.
,
Friederici
,
A. D.
,
Damian
,
M.
,
Meyer
,
A. S.
, &
Levelt
,
W. J. M.
(
2002
).
Semantic category interference in overt picture naming: Sharpening current density localization by PCA
.
Journal of Cognitive Neuroscience
,
14
,
455
462
.
Mahon
,
B. Z.
,
Costa
,
A.
,
Peterson
,
R.
,
Vargas
,
K. A.
, &
Caramazza
,
A.
(
2007
).
Lexical selection is not by competition: A reinterpretation of semantic interference and facilitation effects in the picture–word interference paradigm
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
33
,
503
535
.
Meyer
,
A. S.
(
1990
).
The time course of phonological encoding in language production: The encoding of successive syllables of a word
.
Journal of Memory and Language
,
29
,
524
545
.
Meyer
,
A. S.
(
1991
).
The time course of phonological encoding in language production: Phonological encoding inside a syllable
.
Journal of Memory and Language
,
30
,
69
89
.
Miller
,
J.
,
Patterson
,
T.
, &
Ulrich
,
R.
(
1998
).
Jackknife-based method for measuring LRP onset latency differences
.
Psychophysiology
,
35
,
99
115
.
Miozzo
,
M.
,
Pulvermüller
,
F.
, &
Hauk
,
O.
(
2015
).
Early parallel activation of semantics and phonology in picture naming: Evidence from a multiple linear regression MEG study
.
Cerebral Cortex
,
25
,
3343
3355
.
Mirman
,
D.
(
2014
).
Growth curve analysis and visualization using R
.
Boca Raton, FL
:
Chapman & Hall/CRC Press
.
Munding
,
D.
,
Dubarry
,
A.-S.
, &
Alario
,
F.-X.
(
2016
).
On the cortical dynamics of word production: A review of the MEG evidence
.
Language, Cognition and Neuroscience
,
31
,
441
462
.
Navarrete
,
E.
,
Del Prato
,
P.
, &
Mahon
B. Z.
(
2012
).
Factors determining semantic facilitation and interference in the cyclic naming paradigm
.
Frontiers in Psychology
,
3
,
38
.
Navarrete
,
E.
,
Del Prato
,
P.
,
Peressotti
,
F.
, &
Mahon
,
B. Z.
(
2014
).
Lexical selection is not by competition: Evidence from the blocked naming paradigm
.
Journal of Memory and Language
,
76
,
253
272
.
Oppenheim
,
G. M.
,
Dell
,
G. S.
, &
Schwartz
,
M. F.
(
2010
).
The dark side of incremental learning: A model of cumulative semantic interference during lexical access in speech production
.
Cognition
,
114
,
227
252
.
O'Seaghdha
,
P. G.
,
Chen
,
J.-Y.
, &
Chen
,
T.-M.
(
2010
).
Proximate units in word production: Phonological encoding begins with syllables in Mandarin Chinese but with segments in English
.
Cognition
,
115
,
282
302
.
O'Seaghdha
,
P. G.
, &
Frazer
,
A. K.
(
2014
).
The exception does not rule: Attention constrains form preparation in word production
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
40
,
797
810
.
Perret
,
C.
, &
Laganaro
,
M.
(
2012
).
Comparison of electrophysiological correlates of writing and speaking: A topographic ERP analysis
.
Brain Topography
,
25
,
64
72
.
Peterson
,
R. R.
, &
Savoy
,
P.
(
1998
).
Lexical selection and phonological encoding during language production: Evidence for cascaded processing
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
24
,
539
557
.
Pulvermüller
,
F.
. (
1999
).
Words in the brain's language
.
Behavioral and Brain Sciences
,
22
,
253
279
.
Python
,
G.
,
Fargier
,
R.
, &
Laganaro
,
M.
(
2018
).
ERP evidence of distinct processes underlying semantic facilitation and interference in word production
.
Cortex
,
99
,
1
12
.
Qu
,
Q.
, &
Damian
,
M. F.
(
2020
).
An electrophysiological analysis of the time course of phonological and orthographic encoding in written word production
.
Language, Cognition and Neuroscience
,
35
,
360
373
.
Qu
,
Q.
,
Damian
,
M. F.
, &
Kazanina
,
N.
(
2012
).
Sound-sized segments are significant for Mandarin speakers
.
Proceedings of the National Academy of Sciences, U.S.A.
,
109
,
14265
14270
.
Qu
,
Q.
,
Feng
,
C.
, &
Damian
,
M. F.
(
under review
).
Interference effects of phonological similarity in word production arise from competitive incremental learning
.
Qu
,
Q.
,
Feng
,
C.
,
Hou
,
F.
, &
Damian
,
M. F.
(
2020
).
Syllables and phonemes as planning units in Mandarin Chinese spoken word production: Evidence from ERPs
.
Neuropsychologia
,
146
,
107559
.
Qu
,
Q.
,
Zhang
,
Q.
, &
Damian
,
M. F.
(
2016
).
Tracking the time course of lexical access in orthographic production: An event-related potential study of word frequency effects in written picture naming
.
Brain and Language
,
159
,
118
126
.
Rapp
,
B.
, &
Goldrick
,
M.
(
2000
).
Discreteness and interactivity in spoken word production
.
Psychological Review
,
107
,
460
499
.
Roelofs
,
A.
(
1992
).
A spreading-activation theory of lemma retrieval in speaking
.
Cognition
,
42
,
107
142
.
Roelofs
,
A.
(
1997
).
The WEAVER model of word‐form encoding in speech production
.
Cognition
,
64
,
249
284
.
Roelofs
,
A.
(
2006
).
The influence of spelling on phonological encoding in word reading, object naming, and word generation
.
Psychonomic Bulletin & Review
,
13
,
33
37
.
Roelofs
,
A.
(
2018
).
A unified computational account of cumulative semantic, semantic blocking, and semantic distractor effects in picture naming
.
Cognition
,
172
,
59
72
.
Schmitt
,
B. M.
,
Münte
,
T. F.
, &
Kutas
,
M.
(
2000
).
Electrophysiological estimates of the time course of semantic and phonological encoding during implicit picture naming
.
Psychophysiology
,
37
,
473
484
.
Schnur
,
T. T.
,
Schwartz
,
M. F.
,
Brecher
,
A.
, &
Hodgson
,
C.
(
2006
).
Semantic interference during blocked-cyclic naming: Evidence from aphasia
.
Journal of Memory and Language
,
54
,
199
227
.
Schriefers
,
H.
,
Meyer
,
A. S.
, &
Levelt
,
W. J.
(
1990
).
Exploring the time course of lexical access in language production: Picture–word interference studies
.
Journal of Memory and Language
,
29
,
86
102
.
Starreveld
,
P. A.
, &
La Heij
,
W.
(
1995
).
Semantic interference, orthographic facilitation, and their interaction in naming tasks
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
21
,
686
698
.
Starreveld
,
P. A.
&
La Heij
,
W.
(
1996
).
Time-course analysis of semantic and orthographic context effects in picture naming
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
22
,
896
918
.
Strijkers
,
K.
(
2016a
).
A neural assembly-based view on word production: The bilingual test case
.
Language Learning
,
66
,
92
131
.
Strijkers
,
K.
(
2016b
).
Can hierarchical models display parallel cortical dynamics? A non-hierarchical alternative of brain language theory
.
Language, Cognition and Neuroscience
,
31
,
465
469
.
Strijkers
,
K.
,
Baus
,
C.
,
Runnqvist
,
E.
,
FitzPatrick
,
I.
, &
Costa
,
A.
(
2013
).
The temporal dynamics of first versus second language production
.
Brain and Language
,
127
,
6
11
.
Strijkers
,
K.
, &
Costa
,
A.
(
2011
).
Riding the lexical speedway: A critical review on the time course of lexical selection in speech production
.
Frontiers in Psychology
,
2
,
356
.
Strijkers
,
K.
, &
Costa
,
A.
(
2016
).
The cortical dynamics of speaking: Present shortcomings and future avenues
.
Language, Cognition and Neuroscience
,
31
,
484
503
.
Strijkers
,
K.
,
Costa
,
A.
&
Pulvermüller
,
F.
(
2017
).
The cortical dynamics of speaking: Lexical and phonological knowledge simultaneously recruit the frontal and temporal cortex within 200 ms
.
Neuroimage
,
163
,
206
219
.
Strijkers
,
K.
,
Costa
,
A.
, &
Thierry
,
G.
(
2010
).
Tracking lexical access in speech production: Electrophysiological correlates of word frequency and cognate effects
.
Cerebral Cortex
,
20
,
912
928
.
Taylor
,
J. K.
, &
Burke
,
D. M.
(
2002
).
Asymmetric aging effects on semantic and phonological processes: Naming in the picture–word interference task
.
Psychology and Aging
,
17
,
662
676
.
Thierry
,
G.
,
Cardebat
,
D.
, &
Demonet
,
J.-F.
(
2003
).
Electrophysiological comparison of grammatical processing and semantic processing of single spoken nouns
.
Cognitive Brain Research
,
17
,
535
547
.
Ulrich
,
R.
, &
Miller
,
J.
(
2001
).
Using the jackknife-based scoring method for measuring LRP onset effects in factorial designs
.
Psychophysiology
,
38
,
816
827
.
Van Turennout
,
M.
,
Hagoort
,
P.
, &
Brown
,
C. M.
(
1997
).
Electrophysiological evidence on the time course of semantic and phonological processes in speech production
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
23
,
787
806
.
Wang
,
J.
,
Wong
,
A. W.-K.
,
Wang
,
S.
, &
Chen
,
H.-C.
(
2017
).
Primary phonological planning units in spoken word production are language-specific: Evidence from an ERP study
.
Scientific Reports
,
7
,
5815
.
Wang
,
M.
,
Shao
,
Z.
,
Chen
,
Y.
, &
Schiller
,
N. O.
(
2018
).
Neural correlates of spoken word production in semantic and phonological blocked cyclic naming
.
Language, Cognition and Neuroscience
,
33
,
575
586
.
Wong
,
A. W.-K.
,
Wang
,
J.
,
Ng
,
T.-Y.
, &
Chen
,
H.-C.
(
2016
).
Syllabic encoding during overt speech production in Cantonese: Evidence from temporal brain responses
.
Brain Research
,
1648
,
101
109
.
Yu
,
M.
,
Mo
,
C.
, &
Mo
,
L.
(
2014
).
The role of phoneme in Mandarin Chinese production: Evidence from ERPs
.
PLoS One
,
9
,
e106486
.
Zhang
,
Q.
,
Yu
,
B.
,
Zhang
,
J.
,
Jin
,
Z.
, &
Li
,
L.
(
2018
).
Probing the timing recruitment of Broca's area in speech production for Mandarin Chinese: A TMS study
.
Frontiers in Human Neuroscience
,
12
,
133
.
Zhu
,
X.
,
Damian
,
M. F.
, &
Zhang
,
Q.
(
2015
).
Seriality of semantic and phonological processes during overt speech in Mandarin as revealed by event-related brain potentials
.
Brain and Language
,
144
,
16
25
.