Abstract

The present work investigates the relationship between semantic and prosodic (metric) processing in spoken language under 2 attentional conditions (semantic and metric tasks) by analyzing both behavioral and event-related potential (ERP) data. Participants listened to short sentences ending in semantically and/or metrically congruous or incongruous trisyllabic words. In the metric task, ERP data showed that metrically incongruous words elicited both larger early negative and late positive components than metrically congruous words, thereby demonstrating the online processing of the metric structure of words. Moreover, in the semantic task, metrically incongruous words also elicited an early negative component with similar latency and scalp distribution as the classical N400 component. This finding highlights the automaticity of metrical structure processing. Moreover, it demonstrates that violations of a word's metric structure may hinder lexical access and word comprehension. This interpretation is supported by the behavioral data showing that participants made more errors for semantically congruous but metrically incongruous words when they were attending to the semantic aspects of the sentence. Finally, the finding of larger N400 components to semantically incongruous than congruous words, in both the semantic and metric tasks, suggests that the N400 component reflects automatic aspects of semantic processing.

Introduction

One of the main differences between the written and spoken forms of language is that spoken language conveys prosodic information. Prosody comprises intonation, accentuation, and rhythmic patterns that are produced by variations in acoustic parameters such as fundamental frequency (F0), intensity, duration, and spectral characteristics. Prosody plays both an emotional (Buchanan et al. 2000; Besson et al. 2002; Schirmer et al. 2002) and a linguistic function in spoken language. Here we focused on the latter. The linguistic function of prosody operates both at the lexical level, to facilitate word recognition (e.g., Cutler and Van Donselaar 2001; Soto-Faraco et al. 2001; Copper et al. 2002; Friedrich et al. 2004), and at the structural levels (see Cutler et al. 1997; Hirst and Di Cristo 1998; Cutler and Clifton 1999; Di Cristo 1999 for comprehensive reviews of the functions of prosody). The structural function of prosody in utterance parsing and hierarchical organization can be seen both at the syntactic (e.g., Marslen-Wilson et al. 1992; Beckman 1996; Pynte and Prieur 1996; Warren 1996) and discourse levels (Most and Slatz 1979; Birch and Clifton 1995). Thus, prosody operates at every level of language organization. However, although it is an important feature of language comprehension, prosody has, until recently, received less attention in the psycho- and neurolinguistic literature than other aspects, such as syntax or semantics.

During the past 7 years, several studies have effectively used the event-related potential (ERP) method to study the online processing of prosody during language comprehension by exploring the electrophysiological correlates of different aspects of prosody. At the lexical level, the metrical function of prosody was first examined by Böcker et al. (1999) through the rhythmic properties (metrical stress) of spoken words in Dutch. Participants either listened passively to, or discriminated between, sequences of 4 bisyllabic Dutch words, which started with either weakly (12% of the Dutch lexicon) or strongly stressed syllables (i.e., weak-initial vs. strong-initial words). Results showed that weak-initial words elicited a larger frontal negativity (denoted as N325) than strong-initial words, particularly in the discrimination task. The authors concluded that the N325 may reflect the extraction of metrical stress from the acoustic signal.

Recent results by Friedrich et al. (2004) suggest that F0 also influences word recognition. The authors presented a prime syllable, accented or unaccented, followed by a bisyllabic German word or nonword target which, in half of the trials, started with the same syllable as the prime syllable. The accentuation was materialized through the manipulation of F0. The participants' task was to determine whether the target was a word or a pseudoword. Results demonstrated that reaction times (RTs) were shorter, and the P350 component's amplitude smaller, when the accentual pattern of the bisyllabic target matched with the prime syllable (RE followed by REgel or re followed by reGAL) than when it did not match (e.g., RE followed by reGAL). Thus, accentuation seems to facilitate word recognition by activating the relevant lexical representation. Interestingly, unexpected F0 manipulations of words in sentence contexts also elicit increased positivities in the 300- to 600-ms latency range over the posterior region of the scalp in both adults (Schön et al. 2004) and children (Magne et al. 2006; Moreno and Besson 2006).

At the structural level, a few studies have sought to examine the relationship between prosody, on one hand, and syntax, pragmatics, or semantics, on the other hand, using the ERP method. Steinhauer et al. (1999), for instance, were the first to demonstrate that prosody directs syntactic parsing in the initial analysis of syntactically ambiguous sentences. A positive component closure positive shift (CPS) was elicited by intonation phrase boundaries of cooperating sentences, reflecting the online processing of prosodic structure. Moreover, Pannekamp et al. (2005) used hummed sentences (i.e., without lexical information, but with preserved intonation) to show that prosodic boundaries still elicited CPS components, thus suggesting that the CPS is directly linked to prosodic processing. More recently, Eckstein and Friederici (2005) provided evidence for late interactions between prosodic and syntactic processing by manipulating the position of words that were prosodically marked as sentence-final or sentence-penultimate words in syntactically correct or incorrect sentences. They found that a right anterior negative (RAN) component was elicited by the prosodic manipulations, independently of the syntactic correctness of the sentences.

In order to examine the relation between prosody and pragmatics, and to determine whether listeners make online use of focal prominences/accents to build coherent representations of the informational structure of speech, Magne et al. (2005) used short French dialogues comprised of a question and an answer presented aurally. By manipulating the position of focal accents in the answer, it was possible to render the prosodic patterns either coherent or incoherent with regard to the pragmatic context introduced by the question (e.g., “Did he give his fiancée a ring or a bracelet? He gave a RING to his fiancée” vs. “*He gave a ring to his FIANCEE”; capitalized word bearing a focal accent). Results showed that incoherent prosodic patterns elicited different ERP components depending upon their position within the answer. Although sentence-medial incongruous prosodic patterns elicited a P300-like component that was interpreted as reflecting a prosodic surprise effect, sentence-final incongruous prosodic patterns elicited an N400-like component, possibly reflecting enhanced lexical, semantic, and pragmatic integration difficulties.

Finally, Astésano et al. (2004) investigated the relationship between semantic and prosodic processing by studying the modality function of prosody. Based on the typical findings of increased F0 patterns for questions and decreased F0 patterns for statements (see Hirst and Di Cristo 1998 for a review), prosodic incongruities were created by cross-splicing the beginning of statements with the end of questions, and vice versa. Participants had to decide whether the aurally presented sentences were semantically or prosodically congruous in 2 different attentional conditions (attention to semantics or to prosody). Results showed that a left temporoparietal positive component (P800) was associated with prosodic incongruities, whereas a right centroparietal negative component (N400) was associated with semantic incongruities. Moreover, the P800 component to prosodic incongruities was larger when the sentences were semantically incongruous than congruous, suggesting interactive effects of semantic and prosodic processing. Interestingly, results also showed that the semantic incongruities elicited an N400 component regardless of the orientation of participants' attention (prosody or semantics), thereby suggesting that at least some aspects of semantic processing rely on automatic processes. By contrast, prosodic incongruities elicited a P800 component only when participants focused their attention on the prosodic aspect (modality) of sentences, which may also be linked with the increased difficulty in the detection of prosodic incongruities.

The general aims of the present work were to study the processing of the rhythmic properties of words, through the manipulation of syllabic duration, and the consequences of this manipulation on semantic processing of French spoken sentences. Traditionally, French is described as having fixed accents located at the end of rhythmic groups. This accent is characterized by a lengthening of the last syllable of a word or group of words (Delattre 1966) and contributes to the rhythmic organization of French (Wenk and Wioland 1982; Bailly 1989), as it does in other languages (e.g., Frazier et al. 2004). For instance, Salverda et al. (2003) were able to demonstrate, by recording eye movements, that syllabic lengthening influences lexical interpretation. Items were more likely to be interpreted as monosyllabic words (e.g., ham) than as bisyllabic words (e.g., hamster) when the duration of the first syllable was longer than when it was shorter. Moreover, there is behavioral evidence that French listeners use final syllabic lengthening to speed up detection of a target syllable located at a rhythmic-group boundary in comparison to the same syllable at another location (Dahan 1996). Recently, Christophe et al. (2004) have shown in an elegant study that final lengthening at phonological phrases boundaries facilitates the resolution of local lexical ambiguity. Thus, for instance, although the phonological phrase “un chat drogué” was processed faster than “un chat grincheux,” because of the ambiguity linked with the competitor word “chagrin” in the second example, no difference was found for the phrase “son grand chat grimpait,” supposedly, because the ambiguity straddled the phonological phrase boundary.

The specific aim of the present experiment was 3-fold. First, we wanted to determine whether we could find ERP evidence for the online processing of misplaced stress accents in French. To this end, we created prosodic incongruities by applying the duration of the last syllable to the penultimate syllable of the trisyllabic final words of sentences. We chose to manipulate the second syllable because, according to French metric structure (Astésano 2001), this syllable is never stressed. Based on previous results using different types of prosodic incongruities in sentence or word contexts (Astésano et al. 2004; Friedrich et al. 2004; Schön et al. 2004; Magne et al. 2005, 2006), we predicted that a metric violation of syllabic duration would produce increased late positivities. However, based on the results of Böcker et al. (1999) and Eckstein and Friederici (2005) mentioned above, such an irregular stress pattern may also elicit an increased negativity (e.g., N325 or RAN).

Second, we aimed to determine whether the metric and semantic aspects of spoken language are processed independently or in interaction. Thus, the semantic and metric aspects were manipulated orthogonally, so as to create 4 conditions in which sentences were 1) S+M+, both semantically and metrically congruous, 2) S+M−, semantically congruous and metrically incongruous, 3) S−M+, semantically incongruous and metrically congruous, and finally 4) S−M−, both semantically and metrically incongruous (see Table 1).

Table 1

Examples of stimuli used in the 4 experimental conditions

 Semantically congruous (S+) Semantically incongruous (S−) 
Metrically congruous (M+) Le concours a regroupé mille candidats Le concours a regroupé mille bigoudis 
 “The competition hosted a thousand candidates” “The competition hosted a thousand curlers” 
Metrically incongruous (M−) Le concours a regroupé mille candidats Le concours a regroupé mille bigoudis 
 “The competition hosted a thousand candidates” “The competition hosted a thousand curlers” 
 Semantically congruous (S+) Semantically incongruous (S−) 
Metrically congruous (M+) Le concours a regroupé mille candidats Le concours a regroupé mille bigoudis 
 “The competition hosted a thousand candidates” “The competition hosted a thousand curlers” 
Metrically incongruous (M−) Le concours a regroupé mille candidats Le concours a regroupé mille bigoudis 
 “The competition hosted a thousand candidates” “The competition hosted a thousand curlers” 

Note: The lengthened syllable is underlined.

Finally, in different blocks of trials, participants were asked to focus their attention on the metrical structure or on the semantics of the final words of the sentence to decide whether the final words were metrically congruous or incongruous (metric task) or semantically congruous or incongruous (semantic task). Two questions were of main interest. First, would metric incongruities be associated with similar electrophysiological effects when attention is focused on meter as on semantics? Conversely, would semantic incongruities generate N400 components independently of the direction of attention? This is an important and difficult issue because contradictory results have been reported in the literature. For instance, although Chwilla et al. (1995) found evidence that the N400 component may reflect controlled aspects of the integration of word meaning, Astésano et al. (2004) reported that semantic processing may occur automatically, even when not relevant to the task at hand.

Methods

Participants

Fourteen participants (7 females, mean age 26, age range 23–31), gave their informed consent, and were paid to participate in the experiment, which lasted for about 2 h. All were right-handed native speakers of French, without hearing or neurological disorders.

Stimuli

A total of 512 experimental sentences were built in such a way that they all ended with a trisyllabic noun. Among the 512 sentences, 256 ended with semantically congruous words (S+) and 256 ended with semantically incongruous words (S−, see Table 1). Semantically incongruous sentences were built by replacing the final congruous word with a word that shared the same acoustic and phonological characteristics but that did not make sense in the sentence context. Moreover, semantically congruous and incongruous sentence-final words were matched for word frequency (92.38 and 91.36 occurrences per million, respectively), using the LEXIQUE2 French lexical database (New et al. 2001). Within each semantic condition, half of the sentences ended with an unmodified word with natural lengthening of the last syllable (M+). The other half of the sentences ended with an incongruous lengthening of the penultimate syllable of the final trisyllabic word (M−). This syllabic lengthening was created by increasing the duration of the vowel of the penultimate syllable, using a time-stretching algorithm which allows for the manipulation of the duration of acoustic signals without modifying their timbre or frequency (Pallone 1999 and see below).

Each of the 4 experimental conditions (S+M+, S−M+, S+M−, and S−M−) comprised 32 different sentences (sound examples illustrating stimuli in each experimental condition are available at http://www.lma.cnrs-mrs.fr/∼ystad/CerebralCortex/CerebralCortex.html). Four different experimental lists of 128 sentences were built in order to present each sentence in each experimental condition across subjects, with no repetition within subjects.

Speech Signal

The 256 prosodically congruous sentences were spoken by a native male speaker of standard French and recorded in an anechoic chamber using a digital audiotape (sampling at 44.1 kHz). The mean duration of the sentences was 2.8 s (standard deviation [SD] = 0.9 s), and the mean speech rate was 5.35 syllables per second (SD = 0.65). All sentences were spoken in a declarative mode, and the pitch contour was always falling at the end of the sentence.

For all sentences, the final word comprised 3 syllables and was chosen according to the following constraints. The second and third syllables all possessed a consonant–vowel structure (X-CV-CV, where X could be CV or V). In addition, the second and third syllables never contained a nasal vowel (e.g., forumla]), which are known to have longer durations than the non-nasal vowels in French (e.g., [a], [a], [i], [e], [ε], forumla; Astésano 2001). We also avoided using trisyllabic words with liquid consonants ([l] or [r]) as much as possible, as they are more difficult to segment. All final words were segmented manually by a professional phonetician. The mean duration of final words was 496 ms (SD = 52 ms). On average, the first syllable was 150 ms long (SD = 28 ms), the second syllable was 145 ms long (SD = 28 ms), and the third syllable was 202 ms long (SD = 42 ms). Finally, and as determined from the French lexical database LEXIQUE2 (New et al. 2001), the mean number of phonemes is equal to 6.2, and the mean phonological unicity point (i.e., the rank of the phoneme from which the word can be identified without any ambiguity) is 5.4, which corresponds to an approximate duration of 400 ms (i.e., the unicity point is located, on average, between the onset and offset of the third syllable).

The choice of the lengthening factor that was used to create the metric incongruity was constrained by the necessity of remaining ecological (i.e., to keep the naturalness of speech). The ratio of lengthening for final syllables in French is known to vary considerably as a function of factors such as depth of the adjacent prosodic boundary and speech style (Astésano 2001). In our materials, the lengthening ratio used was equal to 1.7, which is within the range found for natural syllable lengthening of the last syllable in spoken French (1.7–2.5; Astésano 2001). The lengthening of the penultimate syllable creates the auditory impression of a misplaced accented syllable in the words while remaining ecological.

In fact, note that changing the duration of a signal without modifying its frequency is an intricate problem. It was therefore necessary to construct a time-stretching algorithm to modify the syllable length. We decided to use a time-domain approach, in which the signal is time stretched by accurately adding short, nonmodified segments of the original time signal. Consequently, the F0 and amplitude contours of the stretched syllable are identical to the F0 and amplitude contours of the unmodified syllable, but they unfold more slowly over time (i.e., the rate of F0 and amplitude variations differs between the metrically congruous and incongruous conditions; see Fig. 1A,B). The choice of the segments, together with the choice of the position of the insertion, is crucial for the quality of the resulting signal. To ensure that no audible rhythmic defaults or discontinuities occurred, we avoided duplicating the transient part of the signal and constructed segments containing a whole number of periods for periodic signals. To this aim, we built an algorithm, derived from Synchronous Overlap and Add (SOLA) methods (WSOLA and SOLAFS, Dattorro 1987; Laroche 1993), that calculated the ideal duration of the segment to be inserted (see Pallone [1999] for more details on the algorithm used in the present study).

Figure 1.

(A) Original version of the word “canapé.” (B) Time-stretched version of the word “canapé.” On both panels, the waveform is represented on the top part; the spectrogram (gray scale), the F0 (solid line), and the intensity (dashed line) are represented on the bottom part.

Figure 1.

(A) Original version of the word “canapé.” (B) Time-stretched version of the word “canapé.” On both panels, the waveform is represented on the top part; the spectrogram (gray scale), the F0 (solid line), and the intensity (dashed line) are represented on the bottom part.

Because the algorithm stretched the F0 and intensity contours from the beginning of the second syllable, we conducted statistical analyses to determine when these acoustical parameters started to be significantly different between lengthened words and their normal version. For both metrically congruous and metrically incongruous words, F0 and intensity values were extracted using 20 ms steps in between the beginning of the second syllable (considered as time 0 ms) until 100 ms later. Then, the differences between these F0/intensity values and the F0/intensity values at 0 ms were computed. Finally, paired t-tests were conducted between the values for the normal and for the lengthened words. Results revealed that F0 values started to differ significantly between metrically incongruous and metrically congruous words from 80 ms after second syllable onset (t255 = 2.20, P < 0.03) and intensity values from 60 ms after second syllable onset (t255 = 3.28, P < 0.01).

Procedure

Participants were presented with 128 short sentences that were semantically and/or metrically congruous or incongruous. Each experiment began with a practice session to familiarize participants with the task and to train them to blink during the interstimulus interval. The sentences were presented aurally, through headphones, in a pseudorandom order, within 4 blocks of 32 trials each. In 2 blocks, participants were asked to pay attention only to the semantic content in order to decide whether the last word of each sentence was semantically congruous or incongruous. In the other 2 blocks, participants were asked to pay attention only to the syllabic duration in order to decide whether the last word of each sentence was well pronounced or not. Participants were required to press one of 2 buttons as quickly and accurately as possible to give their response. The side (right or left hand) of the response was balanced across participants. Furthermore, half of the participants began with the semantic task and the other half with the metric task.

ERP Recordings

EEG was recorded for 2200 ms, starting 200 ms before the onset of the last word from 28 scalp electrodes, mounted on an elastic cap, and located at standard left and right hemisphere positions over frontal, central, parietal, occipital, and temporal areas (International 10/20 system sites: Fz, Cz, Pz, Oz, Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fc5, Fc1, Fc2, Fc6, Cp5, Cp1, Cp2, Cp6). These recording sites plus an electrode placed on the right mastoid were referenced to the left mastoid electrode. The data were then rereferenced offline to the algebraic average of the left and right mastoids. Impedances of the electrodes never exceeded 3 kΩ. In order to detect horizontal eye movements and blinks, the horizontal electrooculogram (EOG) was recorded from electrodes placed 1 cm to the left and right of the external canthi, and the vertical EOG was recorded from an electrode beneath the right eye, referenced to the left mastoid. Trials containing ocular artifacts, movement artifacts, or amplifier saturation were excluded from the averaged ERP waveforms. The EEG and EOG were amplified by a SA Instrumentation amplifier with a band pass of 0.01–30 Hz and were digitized at 250 Hz by a PC-compatible microcomputer.

Data Analyses

Behavioral data (error rates and RTs) were analyzed using a 3-way analysis of variance (ANOVA), including task (metric or semantic), meter (congruous vs. incongruous), and semantics (congruous vs. incongruous) as within-subject factors. Moreover, mean amplitude ERPs to final words were measured in several latency bands (100–250, 250–450, 500–800, and 800–1200 ms) determined both from visual inspection and from the results of consecutive analyses of 50-ms latency widows. Results were analyzed using ANOVAs that included the same factors as above (task, meter, and semantics) plus electrodes (Fz, Cz, Pz, Oz) for midline analyses and hemispheres (left vs. right), regions of interest (3 ROIs: frontocentral, temporal, and parietotemporal), and electrodes (3 for each ROI: F3, F7, Fc1/F4, F8, Fc2; Fc5, C3, Cp5/Fc6, C4, Cp6; and Cp1, P3, T5/Cp2, P4, T6) for lateral electrodes. When interactions between 2 or more factors were significant, post hoc comparisons between relevant condition pairs were computed. All P values were adjusted with the Greenhouse–Geisser epsilon correction for nonsphericity when necessary.

Results

Behavioral Data

Results of a 3-way ANOVA on the transformed percentages of errors showed no significant main effect of task, meter, or semantics. The meter by semantics interaction was significant (F1,12 = 16.37, P < 0.001): regardless of the direction of attention, participants made more errors when one dimension, meter (19.5%) or semantics (20%), was incongruous than when both dimensions were congruous (12%) or incongruous (16.5%; see Table 2). Finally, the task by meter by semantics interaction was also significant (F1,12 = 4.74, P < 0.05): in the semantic task, participants made more errors when semantics was congruous, but meter was incongruous (S+M−) than in the other 3 conditions.

Table 2

Mean error rates (% Err) in % and mean RTs in ms for each of the 4 experimental conditions (S+M+, S−M+, S+M−, and S−M−) in both the metric and semantic tasks

 Metric Semantic 
 S+M+ S−M+ S+M− S−M− S+M+ S−M+ S+M− S−M− 
% Err 13 (16) 19 (16) 15 (14) 18 (15) 11 (6) 21 (18) 24 (18) 15 (11) 
RTs 932 (152) 1055 (179) 966 (102) 997 (135) 975 (139) 1127 (184) 1011 (164) 1138 (146) 
 Metric Semantic 
 S+M+ S−M+ S+M− S−M− S+M+ S−M+ S+M− S−M− 
% Err 13 (16) 19 (16) 15 (14) 18 (15) 11 (6) 21 (18) 24 (18) 15 (11) 
RTs 932 (152) 1055 (179) 966 (102) 997 (135) 975 (139) 1127 (184) 1011 (164) 1138 (146) 

Note: The SD is indicated in parentheses.

Results of a 3-way ANOVA on the RTs showed a main effect of semantics (F1,12 = 53.70, P < 0.001): RTs were always shorter for semantically congruous (971 ms) than incongruous words (1079 ms; see Table 2). No other effect reached significance.

Electrophysiological Data

Results of the main ANOVAs in the different latency ranges are presented in Table 3. When the main effects or relevant interactions are significant, results of 2 by 2 comparisons are reported in the text.

Table 3

Results of ANOVAs computed on midline and lateral electrodes

Electrodes Factors df Latency windows (ms) 
   100–250 250–450 500–800 800–1200 
   F P F P F P F P 
Midline 1,13 F < 1  F < 1  F < 1  F < 1  
 1,13 F < 1  10.15 0.007 9.14 0.009 F < 1  
 1,13 F < 1  6.17 0.027 2.05 0.17 1.76 0.20 
 T × M 1,13 F < 1  F < 1  9.68 0.008 2.39 0.14 
Lateral 1,13 F < 1  F < 1  F < 1  F < 1  
 1,13 F < 1  4.64 0.05 4.82 0.04 F < 1  
 1,13 F < 1  10.01 0.007 F < 1  F < 1  
 T × M 1,13 F < 1  F < 1  9.47 0.008 4.3 0.06 
 M × H 1,13 F < 1  4.53 0.06 F < 1  F < 1  
Electrodes Factors df Latency windows (ms) 
   100–250 250–450 500–800 800–1200 
   F P F P F P F P 
Midline 1,13 F < 1  F < 1  F < 1  F < 1  
 1,13 F < 1  10.15 0.007 9.14 0.009 F < 1  
 1,13 F < 1  6.17 0.027 2.05 0.17 1.76 0.20 
 T × M 1,13 F < 1  F < 1  9.68 0.008 2.39 0.14 
Lateral 1,13 F < 1  F < 1  F < 1  F < 1  
 1,13 F < 1  4.64 0.05 4.82 0.04 F < 1  
 1,13 F < 1  10.01 0.007 F < 1  F < 1  
 T × M 1,13 F < 1  F < 1  9.47 0.008 4.3 0.06 
 M × H 1,13 F < 1  4.53 0.06 F < 1  F < 1  

Note: df, degrees of freedom; T, task; S, semantic congruity; M, metric congruity; H, hemisphere. Significant effects are marked in bold and marginally significant (P = 0.06) effects in italics. Values for interactions that were never significant in any of the latency bands considered are not reported.

Prior to 250 ms from final word onset, no significant differences were found either at midline or lateral electrodes. In the 250- to 450-ms range, and as can be seen on Figure 2, semantically incongruous words elicited larger negative components than semantically congruous words in both the semantic and the metric tasks and at both midline (difference [d] = −1.53 μV) and lateral electrodes (d = −1.36 μV; main effect of semantics and no task by semantics interaction, see Table 3). This N400 effect was broadly distributed over the scalp (no semantics by electrodes or by hemispheres/ROIs interactions, see Table 3 and Fig. 3). Interestingly, metrically incongruous words also elicited larger negative components than metrically congruous words in both tasks and at both midline (d = −1.13 μV) and lateral electrodes (d = −1.23 μV; main effect of meter and no task by meter interaction, see Table 3 and Fig. 4). Note, however, that the meter by hemisphere interaction was almost significant (P < 0.06): the amplitude of the negative components was somewhat larger over the right hemisphere (metrically congruous vs. incongruous: F1,13 = 15.95, P = 0.001; d = −1.69 μV) than over the left hemisphere (metrically congruous vs. incongruous: F1,13 = 6.04, P = 0.03, d = −1.11 μV; see Fig. 5).

Figure 2.

Averaged electrophysiological data time locked to the onset of semantically congruous (solid line) or semantically incongruous (dashed line) final words, in the metric (A) and semantic tasks (B). Selected traces from 9 electrodes are presented. The latency ranges within which statistical analyses revealed significant effects are shown in gray (250- to 450-ms and 500- to 800-ms latency ranges). In this figure, as in the following ones, the amplitude (in microvolts) is plotted on the ordinate (negative up) and the time (in milliseconds) is on the abscissa.

Figure 2.

Averaged electrophysiological data time locked to the onset of semantically congruous (solid line) or semantically incongruous (dashed line) final words, in the metric (A) and semantic tasks (B). Selected traces from 9 electrodes are presented. The latency ranges within which statistical analyses revealed significant effects are shown in gray (250- to 450-ms and 500- to 800-ms latency ranges). In this figure, as in the following ones, the amplitude (in microvolts) is plotted on the ordinate (negative up) and the time (in milliseconds) is on the abscissa.

Figure 3.

Topographic maps of the semantic congruity effect in the metric (A) and semantic tasks (B). Mean amplitude differences between semantically incongruous and congruous words in the 250- to 450-ms and 500- to 800-ms latency ranges.

Figure 3.

Topographic maps of the semantic congruity effect in the metric (A) and semantic tasks (B). Mean amplitude differences between semantically incongruous and congruous words in the 250- to 450-ms and 500- to 800-ms latency ranges.

Figure 4.

Averaged electrophysiological data time locked to the onset of metrically congruous (solid line) or metrically incongruous (dashed line) final words, in the metric (A) and semantic tasks (B).

Figure 4.

Averaged electrophysiological data time locked to the onset of metrically congruous (solid line) or metrically incongruous (dashed line) final words, in the metric (A) and semantic tasks (B).

Figure 5.

Topographic maps of the metric incongruity effect in the metric (A) and semantic tasks (B). Mean amplitude difference between metrically incongruous and metrically congruous words in the 250- to 450-ms and 500- to 800-ms latency ranges.

Figure 5.

Topographic maps of the metric incongruity effect in the metric (A) and semantic tasks (B). Mean amplitude difference between metrically incongruous and metrically congruous words in the 250- to 450-ms and 500- to 800-ms latency ranges.

In the 500- to 800-ms range, and as can be seen on Figures 2 and 3, semantically incongruous words still elicited relatively larger negativities than semantically congruous words in both the semantic and metric tasks and at both midline (d = −2.73 μV) and lateral electrodes (d = −2.5 μV; main effect of semantics and no task by semantics interaction, see Table 3). By contrast, metrically incongruous words elicited larger positivities than metrically congruous words only in the metric task (no significant main effect of meter but significant task by meter interactions, see Table 3). In the metric task, the metric congruity effect was significant at both midline (d = 3.07 μV; F1,13 = 8.53, P = 0.01) and lateral electrodes (d =2.39 μV; F1,13 = 9.47, P = 0.008, see Figs 4A and 5A).

Finally, in the 800- to 1200-ms range, metrically incongruous words were still somewhat more positive than metrically congruous words only in the metric task (marginally significant task by meter interaction) and at lateral electrodes (d = 2.05 μV; F1,13 = 3.98, P = 0.06, see Table 3 and Fig. 4A).

Discussion

Semantic Congruity Effect

In the semantic task, final semantically incongruous words with respect to the sentence context elicited larger N400 components than semantically congruous words in the 250- to 450-ms latency band. This semantic congruity effect showed a broad distribution over scalp sites, with a slight centroparietal maximum (see Fig. 3B). These differences extended in the 500- to 800-ms latency range with semantically incongruous words still being associated with relatively larger negativities than semantically congruous words. These results are in line with the literature (Kutas and Hillyard 1980; see Kutas and Federmeier 2000; Besson et al. 2004 for recent reviews) and have been interpreted as reflecting the greater difficulties encountered either in integrating semantically incongruous compared with congruous words in ongoing sentence contexts or in generating expectancies for semantically incongruous compared with congruous words, with recent results favoring the latter interpretation (DeLong et al. 2005). This interpretation is also in line with the behavioral data showing longer RTs to semantically incongruous than congruous words. Regarding the time course of the N400 effect, it is interesting to note that, according to acoustic analyses, the mean duration of the final words is 496 ms (SD = 52 ms). Moreover, the phonological unicity point is 5.4 phonemes (mean number of phonemes is 6.2), which corresponds to a word duration of approximately 400 ms (i.e., the isolation point—when the word can be recognized unambiguously—is located between the onset and offset of the third syllable). Clearly, the N400 effect starts before the isolation point, a result in line with the literature (Van Petten et al. 1999, van Berkum et al. 2003, van den Brink et al. 2006). Indeed, van den Brink et al. (2006) have recently shown that the onset of the N400 effect not only occurs prior to the isolation point of sentence-final words but also that the onset of the N400 effect is unaffected by the position (early vs. late) of the isolation point.

Interestingly, semantically incongruous words also elicited larger N400 components than congruous words in the metric task (no task by semantics interaction). Moreover, the scalp distribution of the semantic congruity effect was not significantly different in the metric and semantic tasks; this finding is taken to reflect the similarity of the semantic congruity effect in both tasks (see Fig. 3). Close inspection of Fig. 2 shows that at the F4 electrode, the N400 to semantically incongruous words is larger in the metric than in the semantic task. Because this point is important for the discussion of the automaticity of the N400 component, we computed ANOVAs including F4 only: results showed that the task by semantics interaction was not significant (F1,13 = 1.32, P = 0.27). This finding is taken to reflect the similarity of the semantic congruity effect in both tasks (see Fig. 3). Thus, participants seem to process the meaning of words, even when instructed to focus attention on syllabic duration. These results are in line with those of Astésano et al. (2004) showing the occurrence of N400s to semantic incongruities independently of whether participants focused their attention on the semantic or prosodic aspects (modality contour, interrogative, or declarative) of the sentences.

The finding of a semantic congruity effect when participants were focusing attention on meter seems to argue in favor of the automaticity of semantic processing. This, however, is a complex issue, mainly because it is not clear whether the N400 reflects the automatic or controlled aspects of semantic processing. Indeed, quite mixed evidence can be found in the literature, even when experiments are based on similar designs. For instance, different results have been reported in studies using a semantic priming paradigm, in which a prime word and a target word, which are semantically related or not, are successively presented and listeners perform tasks that require focusing attention on the semantic, lexical, or physical characteristics of words. Besson et al. (1992) found larger N400s to semantically unrelated compared with related visually presented words when the task was to decide whether the prime and target words shared the same first and final letters. This was taken as evidence that the N400 reflects automatic aspects of semantic processing. By contrast, Chwilla et al. (1995) used a semantic priming paradigm and 2 discrimination tasks based on either the semantic aspects (lexical decision task) or the physical aspects (uppercase vs. lowercase letters) of words presented visually. Results showed an N400 priming effect only in the lexical decision task. In the physical task, a P300 effect was found for both related and unrelated targets. Thus, in contrast to the Besson et al. (1992) results above, these results suggest that the N400 effect reflects controlled aspects of semantic processing. This conclusion was in line with the results of Brown and Hagoort (1993). Using a very well controlled masked priming design, they were able to demonstrate evidence for automatic semantic processing in RTs but found no ERP difference between semantically related and unrelated words (no N400 effect). However, Deacon et al. (2000) did recently investigate this issue further. Using a design very similar to the one used by Brown and Hagoort (1993), but within subjects, they did find evidence for semantic priming when participants could not consciously identify the prime words. Deacon et al. (2000) pointed out that because Brown and Hagoort (1993) used a between-subjects design, between-group differences may explain their different findings with masked and unmasked priming. This is in line with other recent results that also did not replicate the findings of Brown and Hagoort (1993) and suggested that automatic processes alone are sufficient to elicit N400 priming effects (Kiefer and Spitzer 2000; Heil et al. 2004). Finally, it should be noted that although the results reported above were found in the visual modality, the present results, together with those of Astésano et al. (2004), were obtained in the auditory modality. This opens the interesting possibility that semantic priming may be more automatic in the auditory than visual modality (Perrin and Garcia-Larrea 2003). This issue will be investigated in further experiments.

Metric Congruity Effect

In the metric task, metrically incongruous words also elicited larger negative components than metrically congruous words in the 250- to 450-ms range (see Fig. 4A), and this metric congruity effect was somewhat larger over the right than the left hemisphere (meter by hemisphere interaction, P < 0.06, see Fig. 5A). These results show that listeners are sensitive to the metric structure of words and perceive, in real time, the unusual but still ecological lengthening of the penultimate syllable. Regarding the time course of the metric congruity effect, it is striking that this effect starts around 100 ms after the onset of the second syllable (around 250 ms after final word onset, see Methods), which is earlier than the offset time of the stretched second syllable. This is to be expected, however, because statistical analyses conducted on F0 and intensity values of the last words revealed that F0 and intensity contours of the stretched second syllable and of the natural unstretched version differed significantly as early as 80 ms and 60 ms, respectively, from the second syllable onset (one could then argue that participants were not sensitive to syllabic duration per se but to the stretched F0 and intensity contours. These aspects are, however, intrinsically linked because a natural slowing down necessarily induces a slowing down in F0 and intensity contours). Thus, because the second syllable varies in duration, F0 and intensity between metrically incongruous and metrically congruous words, the early negative effect may reflect the processing of low-level acoustic factors. However, although this may be the case, negative components have also been reported in the literature when controlling for the influence of such acoustic factors. For instance, in an experiment designed to examine prosodic focus, Magne et al. (2005) found a negativity, with a maximum amplitude around 400 ms (N400), to prosodically incongruous sentence-final words. Because these prosodically incongruous words were prosodically congruous in other discourse contexts, the N400 did not reflect the processing of low-level acoustic factors per se but rather the linguistic relevance of these acoustic cues.

Moreover, as mentioned in the Introduction, Eckstein and Friederici (2005) found a RAN to sentence-final words that were prosodically marked as penultimate words and were consequently prosodically incongruous. Again, Eckstein and Friederici controlled for low-level acoustic factors that, therefore, did not explain the occurrence of the RAN. Thus, the occurrence of the negative components rather seems to reflect the linguistic consequences of unexpected variations in the prosodic features of the words. Finally, the results recently reported by Strelnikov et al. (2006) are important for our understanding of the anatomo-functional basis of prosodic processing. They conducted experiments, using positron emission tomography (PET) and ERPs, that aimed at examining the influence of prosodic cues on sentence segmentation in Russian, a language in which word order is not as informative as it is in English or Romance languages. Interestingly, they found a right frontal negativity to words that preceded an intonational pause whose position in the sentence determined its meaning. Moreover, PET data revealed that the right dorsolateral prefrontal cortex and the right cerebellum were involved in prosodic segmentation. Thus, though it is difficult to infer the neural basis of the right anterior negativities directly from their scalp topography, it is interesting to note that, taken together, the results described above show converging evidence for the implication of right anterior brain structures in the processing of different aspects of prosody (see also Meyer et al. 2002, 2003, 2004).

Interestingly, the metric congruity effect was also significant when participants focused attention on the semantic aspects of the sentences (no task by metric congruity interaction). From these results, it may be concluded that participants did notice that the metric structure of words was incongruous independently of the direction of attention. Thus, as suggested above for the semantic congruity effect, the present results may be taken to reflect the automatic nature of metric processing. Although this may be the case, an alternative and possibly complementary interpretation should also be considered. A close inspection of Figures 4 and 5 indicates that the metric congruity effect shows a right hemisphere lateralization in the metric task but a more bilateral distribution in the semantic task. Such differences in scalp distribution may therefore reflect qualitative differences in the underlying processes. One interesting possibility is that violations of the metric structure interfere with lexical access and thereby hinder access to word meaning. The influence of metric properties of words on their lexical access has been shown in Spanish (Soto-Faraco et al. 2001), German (Friedrich et al. 2004), English (Copper et al. 2002), and Dutch (Cutler et al. 2001). Our present results show that metric properties also influence lexical access in French. Moreover, they suggest that such integration problems would be reflected by typically bilateral N400 components. Such an interpretation is in line with the behavioral data showing that semantically congruous but metrically incongruous words produced the highest error rates in the semantic task (see Table 2), as if metric incongruities disrupted semantic processing.

At present, we can only speculate about the possible mechanisms responsible for such effects. It may be that participants processed the metrically violated trisyllabic words as if they were metrically correct bisyllabic words with final syllabic lengthening, as shown by Salverda et al. (2003) for mono- and bisyllabic words. An N400 would be generated because such bisyllabic patterns are not real French words. An alternative interpretation is that the N400 is elicited by a prosodic mismatch between the expectation of finality caused by the lengthened second syllable and the following unexpected third syllable. More generally, and based on spoken word recognition models (e.g., Marslen-Wilson and Welsh 1978; McClelland and Elman 1986; McQueen et al. 1994), one may consider that none of the lexical candidates that were activated on the basis of the acoustic properties of the word fit with the violated metrical structure that was heard, thereby hindering access to word meaning and increasing integration costs. In any event, the most important point may be that results of several experiments converge in showing that syllabic lengthening (Salverda et al. 2003; Eckstein and Friederici 2005), correct accentuation (Cutler et al. 2001; Soto-Faraco et al. 2001; Copper et al. 2002; Friedrich et al. 2004), and initial phonological overlap (Connolly and Philips 1994; Van Petten et al. 1999; van den Brink et al. 2001; van den Brink and Hagoort 2004) influence lexical interpretation and lexical access.

Metrically incongruous words also elicited larger late positive components (although late positivities differ in amplitude across conditions, they were nevertheless always present. This is to be expected insofar as participants were asked to make a decision on the sentence final words [Kutas et al. 1977; McCarthy and Donchin 1979; Ragot and Renault 1981; Donchin and Coles 1988]) than metrically congruous words, but only in the metric task (see Figs 4A and 5A). These positivities developed in the 500- to 800-ms latency band and were still significant in the 800- to 1200-ms range. They were broadly distributed across scalp sites. This finding is in line with previous ones showing that the manipulation of different acoustic parameters of the speech signal, such as F0 and intensity, is also associated with increased positivity (Astésano et al. 2004; Friedrich et al. 2004; Schön et al. 2004; Eckstein and Friederici 2005; Magne et al. 2005, 2006; Moreno and Besson 2006). It has been proposed that these late positivities may belong to the P300 family of components elicited by surprising and task-relevant events (Donchin 1981; Donchin and Coles 1988). Indeed, manipulations of acoustic properties of the speech signal were unexpected and may have therefore produced a prosodic surprise effect in the listeners. Moreover, late positivities in sentence contexts are interpreted as reflecting the integration of syntactic, semantic, and pragmatic information (e.g., Kaan et al. 2000). The enhanced positivity reported here may therefore reflect the integration difficulties resulting from the violation of the typical metric pattern of final sentence words in French. Interestingly, no such increase in positivity to metrically incongruous words was observed when participants focused their attention on the semantics of the word (see Figs 4B and 5B). These results are in line with those reported by Astésano et al. (2004) and showing that violations of the intonation pattern of interrogative and declarative sentences elicit larger late positivities only when participants focused their attention on prosody. Thus, both types of prosodic violations (metric and modality) are associated with increased late positivities only when relevant to the task at hand.

Summary and Conclusion

In accordance with previous findings (e.g., Böcker et al. 1999; Salverda et al. 2003; Eckstein and Friederici 2005), the present results show evidence for the online processing of misplaced stress accents in French, through the occurrence of increased early negative and late positive components to metric incongruities in the metric task. In addition, in the semantic task, metrically incongruous words also elicited a negative component that may be considered as a member of the N400 family due to its N400-like latency, amplitude, and scalp distribution. This interpretation is supported by the present behavioral data. In particular, significant interactions found in statistical analyses of error rates showed that participants often (mistakenly) judged the metrically incongruous words as semantically incongruous even when they were attending to semantics and thus instructed to ignore the meter. Furthermore, previous findings showing enhanced negativities to prosodic incongruities when basic acoustic features of stimuli are controlled (e.g., Magne et al. 2005, Eckstein and Friederici 2005) lead us to believe that violations of prosodic expectancies, in this case inappropriately lengthened syllables, are manifested in negative ERP components that do not simply reflect differences in acoustic input but also represent significant interactions with lexical and/or semantic processes. Therefore, the present results highlight the importance of the metric structure of words for language comprehension in French. Finally, the similarity of the semantic congruity effects in both the semantic and metric tasks provides additional evidence for the N400 component's sensitivity to the automatic aspect of semantic processing.

This research was supported by a grant from the Human Frontier Science Program (HSFP #RGP0053) to Mireille Besson. Cyrille Magne benefited from a Cognitive Science fellowship from the French Ministry of Research, and Corine Astésano was a postdoctorate researcher supported by a grant from HFSP. The authors gratefully acknowledge Mario Rossi, Anne Lacheret, Michel Morel, and Valerie Pasdeloup for stimulating discussions, Monique Chiambretto and Reyna Leigh Gordon for their technical assistance, and 3 anonymous reviewers for their precious comments. Conflict of Interest: None declared.

References

Astésano
C
Rythme et accentuation en français
Invariance et variabilité stylistique
 , 
2001
Paris
L'Harmattan
Astésano
C
Besson
M
Alter
K
Brain potentials during semantic and prosodic processing in French
Brain Res Cogn Brain Res
 , 
2004
, vol. 
18
 (pg. 
172
-
184
)
Bailly
G
Integration of rhythmic and syntactic constraints in a model of generation of French prosody
Speech Comm
 , 
1989
, vol. 
8
 (pg. 
137
-
146
)
Beckman
ME
The parsing of prosody
Lang Cogn Process
 , 
1996
, vol. 
11
 (pg. 
17
-
67
)
Besson
M
Fischler
I
Boaz
T
Raney
G
Effects of automatic associative activation on explicit and implicit memory tests
J Exp Psychol Learn Mem Cogn
 , 
1992
, vol. 
18
 (pg. 
89
-
105
)
Besson
M
Magne
C
Regnault
P
Renault
B
Le traitement du langage
L'imagerie fonctionnelle électrique (EEG) et magnétique (MEG): Ses applications en sciences cognitives
 , 
2004
Paris
Hermés
(pg. 
185
-
216
)
Besson
M
Magne
C
Schön
D
Emotional Prosody: sex differences in sensitivity to speech melody
Trends Cogn Sci
 , 
2002
, vol. 
6
 (pg. 
405
-
407
)
Birch
S
Clifton
CE
Focus, accent, and argument structure: effects on language comprehension
Lang Speech
 , 
1995
, vol. 
38
 (pg. 
365
-
391
)
Böcker
KBE
Bastiaansen
MCM
Vroomen
J
Brunia
CHM
De Gelder
B
An ERP correlate of metrical stress in spoken word recognition
Psychophysiology
 , 
1999
, vol. 
36
 (pg. 
706
-
720
)
Brown
CM
Hagoort
P
The processing nature of the N400: evidence from masked priming
J Cogn Neurosci
 , 
1993
, vol. 
5
 (pg. 
34
-
44
)
Buchanan
TW
Lutz
K
Mirzazade
S
Specht
K
Shah
NJ
Zilles
K
Jancke
J
Recognition of emotional prosody and verbal components of spoken language: an fMRI study
Brain Res Cogn Brain Res
 , 
2000
, vol. 
9
 (pg. 
227
-
238
)
Christophe
A
Peperkamp
S
Pallier
C
Block
E
Mehler
J
Phonological phrase boundaries constrain lexical access: I. adult data
J Mem Lang
 , 
2004
, vol. 
51
 (pg. 
523
-
547
)
Chwilla
DJ
Brown
CM
Hagoort
P
The N400 as a function of the level of processing
Psychophysiology
 , 
1995
, vol. 
32
 (pg. 
274
-
285
)
Connolly
JF
Phillips
NA
Event-related potential components reflect phonological and semantic processing of the terminal word of spoken sentences
J Cogn Neurosci
 , 
1994
, vol. 
6
 (pg. 
256
-
266
)
Copper
N
Cutler
A
Wales
R
Constraints of lexical stress on lexical access in English: evidence from native and non-native listeners
Lang Speech
 , 
2002
, vol. 
45
 (pg. 
207
-
228
)
Cutler
A
Clifton
C
Brown
CM
Hagoort
P
Comprehending spoken language: a blueprint of the listener
The neurocognition of language
 , 
1999
Oxford
Oxford University Press
(pg. 
123
-
166
)
Cutler
A
Dahan
D
van Donselaar
W
Prosody in the comprehension of spoken language: a literature review
Lang Speech
 , 
1997
, vol. 
40
 (pg. 
141
-
201
)
Cutler
A
Van Donselaar
WA
Voornaam is not (really) a homophone: Lexical prosody and lexical access in Dutch
Language and Speech
 , 
2001
, vol. 
44
 (pg. 
171
-
195
)
Dahan
D
The role of rhythmic groups in the segmentation of continuous French speech
 , 
1996
 
Proceedings of the Fourth International Conference of Speech and Language Processing; 1996 Oct 3-6; Philadelphia. p. 1185–1188
Dattorro
J
Using digital signal processor chips in a stereo audio time compressor/expander
Journal of the Audio Engineering Society
 , 
1987
, vol. 
35
 pg. 
1062
 
Deacon
D
Hewitt
S
Yang
C
Nagata
M
Event-related potential indices of semantic priming using masked and unmasked words: evidence that the N400 does not reflect a post-lexical process
Brain Res Cogn Brain Res
 , 
2000
, vol. 
9
 (pg. 
137
-
146
)
Delattre
P
Studies in French and comparative phonetics
 , 
1966
The Hague (The Netherlands)
Mouton
DeLong
KA
Urbach
TP
Kutas
M
Probabilistic word pre-activation during language comprehension inferred from electrical brain activity
Nat Neurosci
 , 
2005
, vol. 
8
 (pg. 
1117
-
1121
)
Di Cristo
A
Le cadre accentuel du français contemporain: essai de modélisation (1ère partie)
Langues
 , 
1999
, vol. 
2
 (pg. 
184
-
205
)
Donchin
E
Surprise!…Surprise?
Psychophysiology
 , 
1981
, vol. 
18
 (pg. 
493
-
513
)
Donchin
E
Coles
MGH
Is the P300 component a manifestation of context-updating?
Behav Brain Sci
 , 
1988
, vol. 
11
 (pg. 
355
-
372
)
Eckstein
K
Friederici
AD
Late interaction of syntactic and prosodic processes in sentence comprehension as revealed by ERPs
Brain Res Cogn Brain Res
 , 
2005
, vol. 
25
 (pg. 
130
-
143
)
Frazier
L
Clifton
C
Carlson
K
Don't break, or do: prosodic boundary preferences
Lingua
 , 
2004
, vol. 
114
 (pg. 
3
-
27
)
Friedrich
CK
Kotz
SA
Friederici
AD
Alter
K
Pitch modulates lexical identification in spoken word recognition: ERP and behavioral evidence
Brain Res Cogn Brain Res
 , 
2004
, vol. 
20
 (pg. 
300
-
308
)
Heil
M
Rolke
B
Pecchinenda
A
Automatic semantic activation is no myth: N400 semantic priming effects in the letter search task in the absence of RT effects
Psychol Sci
 , 
2004
, vol. 
15
 (pg. 
852
-
885
)
Hirst
DJ
Di Cristo
A
Hirst
DJ
Di Cristo
A
A survey of intonation systems
Intonation systems: a survey of twenty languages
 , 
1998
Cambridge (UK)
Cambridge University Press
(pg. 
1
-
44
)
Kaan
E
Harris
A
Gibson
E
Holcomb
PJ
The P600 as an index of syntactic integration difficulty
Lang Cogn Process
 , 
2000
, vol. 
15
 (pg. 
159
-
201
)
Kiefer
M
Spitzer
M
Time course of conscious and unconscious semantic brain activation
Neuroreport
 , 
2000
, vol. 
11
 (pg. 
2401
-
2407
)
Kutas
M
Federmeier
KD
Electrophysiology reveals semantic memory use in language comprehension
Trends Cogn Sci
 , 
2000
, vol. 
4
 (pg. 
463
-
470
)
Kutas
M
Hillyard
SA
Reading senseless sentences: brain potentials reflect semantic incongruity
Science
 , 
1980
, vol. 
207
 (pg. 
203
-
205
)
Kutas
M
McCarthy
G
Donchin
E
Augmenting mental chronometry: the P300 as a measure of stimulus evaluation time
Science
 , 
1977
, vol. 
197
 (pg. 
792
-
795
)
Laroche
J
Autocorrelation method for high quality time/pitch scaling
In: Proceedings of IEEE Workshop on Applications of Signal Processing to Audio and Acoustics
 , 
1993
New York
New Paltz
 
Conference date: 17-20 Oct 1993
Magne
C
Astésano
C
Lacheret-Dujour
A
Morel
M
Besson
M
On-line processing of “pop-out” words in spoken French dialogues
J Cogn Neurosci
 , 
2005
, vol. 
17
 (pg. 
740
-
756
)
Magne
C
Schön
D
Besson
M
Musician children detect pitch violations in both music and language better than non-musician children: behavioral and electrophysiological approaches
J Cogn Neurosci
 , 
2006
, vol. 
18
 (pg. 
199
-
211
)
Marslen-Wilson
WD
Tyler
LK
Warren
P
Grenier
P
Lee
CS
Prosodic effects in minimal attachment
Q J Exp Psychol
 , 
1992
, vol. 
45A
 (pg. 
73
-
87
)
Marslen-Wilson
WD
Welsh
A
Processing interactions during word recognition in continuous speech
Cogn Psychol
 , 
1978
, vol. 
10
 (pg. 
29
-
63
)
McCarthy
G
Donchin
E
A metric for thought: a comparison of P300 latency and reaction times
Science
 , 
1979
, vol. 
211
 (pg. 
77
-
80
)
McClelland
JL
Elman
JL
The TRACE model of speech perception
Cogn Psychol
 , 
1986
, vol. 
18
 (pg. 
1
-
86
)
McQueen
JM
Norris
D
Cutler
A
Competition in spoken word recognition: spotting words in other words
J Exp Psychol Learn Mem Cogn
 , 
1994
, vol. 
20
 (pg. 
621
-
638
)
Meyer
M
Alter
K
Friederici
AD
Towards the cerebral substrates of sentence-level syntactic and prosodic processing
J Neurolinguistics
 , 
2003
, vol. 
16
 (pg. 
277
-
300
)
Meyer
M
Alter
K
Friederici
AD
Lohmann
G
von Cramon
Y
FMRI reveals brain regions mediating slow prosodic modulations in spoken sentences
Hum Brain Mapp
 , 
2002
, vol. 
17
 (pg. 
73
-
88
)
Meyer
M
Steinhauer
K
Alter
K
Friederici
AD
von Cramon
DY
Brain activity varies with modulation of dynamic pitch variance in sentence melody
Brain Lang
 , 
2004
, vol. 
89
 (pg. 
277
-
289
)
Moreno
S
Besson
M
Musical training and language-related brain electrical activity in children
Psychophysiology
 , 
2006
, vol. 
43
 (pg. 
287
-
291
)
Most
RB
Saltz
E
Information structure in sentences: new information
Lang Speech
 , 
1979
, vol. 
22
 (pg. 
89
-
95
)
New
B
Pallier
C
Ferrand
L
Matos
R
Une base de données lexicales du français contemporain sur Internet: LEXIQUE
Année Psychologique
 , 
2001
, vol. 
101
 (pg. 
447
-
462
)
Pallone
G
Boussard
P
Daudet
L
Guillemain
P
Kronland-Martinet
R
A wavelet based method for audio-video synchronization in broadcasting applications
 , 
1999
 
In: Proceedings of the Conference on Digital Audio Effects (DAFx-99); 1999 Dec 9–11; Trondheim, Norway. p. 59–62
Pannekamp
A
Toepel
U
Alter
K
Hahne
A
Friederici
AD
Prosody-driven sentence processing: an ERP study
J Cogn Neurosci
 , 
2005
, vol. 
17
 (pg. 
407
-
421
)
Perrin
F
Garcia-Larrea
L
Modulation of the N400 potential during auditory phonological/semantic interaction
Brain Res Cogn Brain Res
 , 
2003
, vol. 
17
 (pg. 
36
-
47
)
Pynte
J
Prieur
B
Prosodic breaks and attachment decisions in sentence parsing
Lang Cogn Process
 , 
1996
, vol. 
11
 (pg. 
165
-
191
)
Ragot
R
Renault
B
P300 as a function of S-R compatibility and motor programming
Biol Psychol
 , 
1981
, vol. 
13
 (pg. 
289
-
294
)
Salverda
AP
Dahan
D
McQueen
S
The role of prosodic boundaries in the resolution of lexical embedding in speech comprehension
Cognition
 , 
2003
, vol. 
90
 (pg. 
51
-
89
)
Schirmer
A
Kotz
SA
Friederici
AD
Sex differentiates the role of emotional prosody during word processing
Brain Res Cogn Brain Res
 , 
2002
, vol. 
14
 (pg. 
228
-
233
)
Schön
D
Magne
C
Besson
M
The music of speech: electrophysiological study of pitch perception in language and music
Psychophysiology
 , 
2004
, vol. 
41
 (pg. 
341
-
349
)
Soto-Faraco
S
Sebastian-Galles
N
Cutler
A
Segmental and suprasegmental mismatch in lexical access
J Mem Lang
 , 
2001
, vol. 
45
 (pg. 
412
-
432
)
Steinhauer
K
Alter
K
Friederici
AD
Brain potentials indicate immediate use of prosodic cues in natural speech processing
Nat Neurosci
 , 
1999
, vol. 
2
 (pg. 
191
-
196
)
Strelnikov
KN
Vorobyev
VA
Chernigovskaya
TV
Medvedev
SV
Prosodic clues to syntactic processing—a PET and ERP study
Neuroimage
 , 
2006
, vol. 
29
 (pg. 
1127
-
1134
)
van Berkum
JJ
Zwitserlood
P
Hagoort
P
Brown
CM
When and how do listeners relate a sentence to the wider discourse? Evidence from the N400 effect
Brain Res Cogn Brain Res
 , 
2003
, vol. 
17
 (pg. 
701
-
718
)
van den Brink
D
Brown
CM
Hagoort
P
Electrophysiological evidence for early contextual influences during spoken-word recognition: N200 versus N400 effects
J Cogn Neurosci
 , 
2001
, vol. 
13
 (pg. 
967
-
985
)
van den Brink
D
Brown
CM
Hagoort
The cascaded nature of lexical selection and integration in auditory sentence processing
J Exp Psychol Learn Mem Cogn
 , 
2006
, vol. 
32
 (pg. 
364
-
372
)
van den Brink
D
Hagoort
P
The influence of semantic and syntactic context constraints on lexical selection and integration in spoken-word comprehension as revealed by ERPs
J Cogn Neurosci
 , 
2004
, vol. 
16
 (pg. 
1068
-
1084
)
Van Petten
C
Coulson
S
Rubin
S
Plante
E
Parks
M
Time course of word identification and semantic integration in spoken language
J Exp Psychol Learn Mem Cogn
 , 
1999
, vol. 
25
 (pg. 
394
-
417
)
Warren
P
Prosody and parsing: an introduction
Lang Cogn Process
 , 
1996
, vol. 
11
 (pg. 
1
-
16
)
Wenk
BJ
Wioland
F
Is French really syllabletimed?
J Phonetics
 , 
1982
, vol. 
10
 (pg. 
193
-
216
)