Distinguishable neural correlates of verbs and nouns: a MEG study on homonyms.

Neuropsychologia 54 (2014) 87–97

Contents lists available at ScienceDirect

Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia

Distinguishable neural correlates of verbs and nouns: A MEG study on homonyms Styliani Tsigka a, Christos Papadelis a, Christoph Braun a,b,c, Gabriele Miceli a,d,n a

CIMeC, Center for Mind/Brain Sciences, University of Trento, Via delle Regole 101, 38100 Trento, Italy MEG-Center, University of Tübingen, Otfried-Müller-Str. 47, 72076 Tübingen, Germany c DiPSCo, Corso Bettini 31, 38068 Rovereto, Italy d CeRiN, Center for Neurocognitive Rehabilitation, CIMeC, Center for Mind/Brain Sciences, University of Trento, Via Matteo del Ben, 5/b, 38068 Rovereto, TN, Italy b

art ic l e i nf o

a b s t r a c t

Article history: Received 24 December 2012 Received in revised form 23 November 2013 Accepted 19 December 2013 Available online 31 December 2013

The dissociability of nouns and verbs and of their morphosyntactic operations has been firmly established by lesion data. However, the hypothesis that they are processed by distinct neural substrates is inconsistently supported by neuroimaging studies. We tackled this issue in a silent reading experiment during MEG. Participants silently read noun/verb homonyms in minimal syntactic context: article-noun (NPs), pronoun-verb (VPs) (e.g., il ballo/i balli, the dance/the dances; io ballo/tu balli, I dance/you dance). Homonyms allow to rule out prelexical or postlexical nuisance factors—they are orthographically and phonologically identical, but serve different grammatical functions depending on context. Under these experimental conditions, different activity to nouns and verbs can be confidently attributed to representational/processing distinctions. At the sensor level, three components of event-related magnetic fields were observed for the function word and four for the content word, but Global Field Power (GFP) analysis only showed differences between VPs and NPs at several but very short time windows. By contrast, source level analysis based on Minimum Norm Estimates (MNE) yielded significantly greater activity for VPs in left frontal areas and in a left frontoparietal network at late time windows (380–397 and 393–409 ms). These results are fully consistent with lesion data, and show that verbs and nouns are processed differently in the brain. Frontal and parietal activation to verbs might correspond to morphosyntactic processes and to working memory recruitment (or thematic role assignment), respectively. Findings are consistent with the view that nouns and verbs and their morphosyntactic operations involve at least partially distinct neural substrates. However, they do not entirely rule out that nouns and verbs are processed in a shared neural substrate, and that differences result from greater complexity of verbal morphosyntax. & 2013 Elsevier Ltd. All rights reserved.

Keywords: Magnetoencephalography Verb processing Noun processing Left inferior Prefrontal cortex Posterior parietal cortex Morphosyntactic processing Verbal working memory Thematic role assignment

1. Introduction The dissociability between the processing of nouns and verbs has been extensively documented in aphasiological data, and in a variety of tasks (Caramazza & Hillis, 1991; Damasio & Tranel, 1993; Miceli, Silveri, Villa, & Caramazza, 1984; Shapiro & Caramazza, 2003; Tsapkini, Jarema, & Kehayia, 2002). The most robust anatomical findings associate impaired verb and noun processing with damage to left frontal and left temporal regions, respectively (e.g. Bak, O0 Donovan, Xuereb, Boniface, & Hodges, 2001; Damasio & Tranel, 1993; Daniele, Giustolisi, Silveri, Colosimo, & Gainotti, 1994). n Corresponding author at: University of Trento, Center for Neurocognitive Rehabilitation, Center for Mind/Brain Sciences, Via Matteo del Ben, 5/b, 38068 Rovereto, TN, Italy. Tel.: þ 39 464 808155; fax: þ 39 464 808150. E-mail address: [email protected] (G. Miceli).

0028-3932/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.neuropsychologia.2013.12.018

Even though the functional dissociability and its most typical neural correlates have been repeatedly demonstrated in braindamaged patients, the search for its neural underpinnings via neuroimaging techniques has yielded inconsistent results. Some PET studies showed left prefrontal and middle frontal cortices to be selectively activated by verbs (Petersen, Fox, Posner, Mintum, & Raichle, 1989; Raichle et al., 1994; Wise et al., 1991). However, other PET studies failed to identify differential neuronal substrates for nouns and verbs (e.g. Tyler, Russell, Fadili, & Moss, 2001; Warburton et al., 1996), even though in some cases greater activation to verbs than nouns was observed (Perani et al., 1999). A large variability exists also in the fMRI literature. While some investigations show greater activation for verbs in the left inferior prefrontal cortex (e.g. Davis, Meunier, & Marslen-Wilson, 2004; Finocchiaro, Basso, Giovenzana, & Caramazza, 2010; Shapiro, Moo, & Caramazza, 2006; Yokoyama et al., 2006), others report on contrasting results. Thus, while in some cases verb processing was

88

S. Tsigka et al. / Neuropsychologia 54 (2014) 87–97

linked to left middle temporal gyrus activation (Longe, Randall, Stamatakis, & Tyler, 2007; Tyler, Randall, & Stamatakis, 2008; Bedny, Caramazza, Grossman, Pascual-Leone, & Saxe, 2008), in others it highlighted a very extensive neural network (Berlingeri et al., 2008), or the same circumscribed brain regions as nouns (Siri et al., 2008). In the abovementioned studies, activation in response to nouns was either unidentifiable (Longe et al., 2007; Tyler et al., 2008) or less robust than to verbs (Berlingeri et al., 2008; Siri et al., 2008). And, in a recent study, noun phrases activated left inferior prefrontal regions more than verbs (Pulvermüller, Cook, & Hauk, 2012). Contrasting results prompted the proposal that the networks underlying noun and verb processing are not spatially segregated (e.g. Crepaldi, Berlingeri, Paulesu, & Luzzatti, 2011; Vigliocco, Vinson, Druks, Barber, & Cappa, 2011). Repetitive Transcranial Magnetic Stimulation (rTMS) and event related potential (ERP) studies also documented distinctions between grammatical categories, but diverged when identifying the neural underpinnings of verb processing. While in a study stimulation of the left prefrontal cortex disrupted verb production (Shapiro, Pascual-Leone, Mottaghy, Gangitano, & Caramazza, 2001), another study failed to confirm this finding (Cappa, Sandrini, Rossini, Sosta, & Miniussi, 2002); and yet another identified the anterior portion of the left middle frontal gyrus as the critical area for inflected verb production (Cappelletti, Fregni, Shapiro, Pascual-Leone, & Caramazza, 2008). ERP studies repeatedly showed increased left-lateralized anterior positivity associated with the processing of verbs as compared to nouns (Dehaene, 1995; Federmeier, Segal, Lombrozo, & Kutas, 2000). However, while different temporal patterns of activation for nouns and verbs are a consistent finding in the ERP literature, topographical differences between the two grammatical categories are variable (Gomes, Ritter, Tartter, Vaughan, & Rosen, 1997; Khader & Rösler, 2004; Pulvermüller, Preißl, Lutzenberger, & Birbaumer, 1996; Pulvermüller, Preißl, & Lutzenberger, 1999). In another study, ERP activity was affected both by grammatical class and by semantic properties (Barber, Kousta, Otten, & Vigliocco, 2010). We used magnetoencephalography (MEG), that offers a potentially more promising approach to the problem under investigation than either fMRI/PET, which rely on slow metabolic changes and therefore have limited temporal resolution, or ERP, whose topographical information is rather underspecified and therefore allows limited claims in terms of separation and localization of underlying generators. MEG combines excellent temporal resolution with good localization accuracy, at least for superficial cortical sources (Papadelis, Poghosyan, Fenwick, & Ioannides, 2009). Therefore, it allows to adequately tackle our primary questions—establishing if nouns and verbs activate identical or distinct neural substrates, and if they do so to the same extent and at the same or different points in time. Available MEG findings on noun/verb processing are controversial. In picture naming studies, Sörös, Cornelissen, Laine, and Salmelin (2003) found identical patterns of activation for both word types, but Liljeström, Hultén, Parkkonen, and Salmelin (2009) observed differences only at an early time window (100– 200 ms). In this latter study, activation of right frontal and bilateral parietal cortex was enhanced by nouns; the anterior-superior temporal lobe was activated by verbs, but weakly and irregularly across subjects. Observations from silent reading are also inconsistent. In Xiang and Xiao (2009), the same regions were activated by nouns and verbs at an early stage, and spatiotemporal sequences diverged at late latencies. In a category judgment task, Fiebach, Maess, and Friederici (2002) examined the effects of syntactic context. When nouns and verbs were presented in isolation, no differences were found in the left hemisphere. However, when they were presented in a minimal syntactic context, nouns elicited stronger magnetic fields over left posterior temporal regions.

Inconsistent MEG results might be due to different causes. Picture naming paradigms are problematic, as noun and verb stimuli typically differ in visual complexity and require different processes for response elicitation (nouns are on average more referential, concrete and imageable). For example, a picture of scissors suffices to precipitate naming of the object, but in order to elicit “to cut”, both scissors and something being cut must be shown. In addition, naming a target verb requires more than analyzing the physical features of the stimulus—saying “to fall” in response to a picture requires not only accurate visuoperceptual analysis, but also assumptions on events that take place before and after the instant captured in the stimulus. These nuisance factors may elicit different activations, independent of grammatical class distinctions. At face value, silent reading tasks are less problematic, as written nouns and verbs can be matched for psycholinguistic variables (length, frequency of usage, etc.). However, also silent reading of unambiguous nouns and verbs (Xiang & Xiao, 2009; Fiebach et al., 2002) is not problem-free, as selected stimuli visually and orthographically different, and the corresponding covert responses may activate different phonological word forms. Given the sensitivity of MEG to even minor changes in visual stimulus features (e.g., Ramkumar, Jas, Pannasch, Hari, & Parkkonen, 2013), also in this case nuisance factors may interfere with results, and render their interpretation problematic. To overcome these shortcomings, we presented homonyms in a silent reading paradigm, to native speakers of Italian. Homonyms were selected because they are orthographically and phonologically identical1, even though they serve different grammatical functions, depending on syntactic context. This choice ensures that during the experimental procedure visuo-perceptual, orthographic, phonological and subvocal processes are engaged to exactly the same extent by nouns and verbs, thus eliminating critical nuisance factors. Under these experimental conditions, different MEG responses to NPs and VPs can be legitimately ascribed to genuine representational/processing (and neural) distinctions between nouns and verbs. An additional advantage afforded by Italian homonyms is that for the most part they are strictly related semantically (e.g., ballo, the dance/I dance; canti, the songs/you sing; cena, the dinner/he has dinner). Therefore, results allow to address a debated issue—whether putative differences between nouns and verbs are the result of distinctions at the lexical–grammatical or at the semantic level (for contrasting views, see Shapiro & Caramazza, 2003; Vigliocco et al., 2011). Neuromagnetic brain responses were analyzed by using the Minimum-Norm Estimates (MNE) (Hämäläinen & Ilmoniemi, 1984, 1994; Hauk, 2004). Results help identify the neural underpinnings of NP and VP processing, and allow discussing some mechanisms potentially underlying noun/verb dissociations.

2. Materials and methods 2.1. Participants Thirteen healthy native Italian speakers participated in this study. They were all right-handed, with normal or corrected-to-normal vision. None reported a history of significant head injury or neurological disease. Prior to testing, written informed consent was obtained from each participant. Compensation was given for participation, following completion of the experiment. The research protocol was approved by the local ethical committee and the study complied with the Declaration of Helsinki. MEG recordings of 12 participants (age: 23–34, mean age: 27; five female and seven male) entered the analysis of the present study. Data from a participant were excluded due to heavy artifact contamination.

1 Owing to the transparency of the relationships between orthography and pronunciation in Italian, words that are homographs are also homophones, with an extremely limited number of exceptions.


2.2. Selection of stimuli A large number of homonyms were extracted from the ‘itWaC0 corpus (Baroni, Bernardini, Ferraresi, & Zanchetta, 2009). When preparing the experimental list, the first selection criterion was that identical word forms be available for the first person singular of the verb and for the singular of the noun (e.g. io canto, I sing; il canto; the song; io ballo, I dance; il ballo, the dance), and for the second person singular of the verb and the plural of the noun (e.g., tu canti, you sing; i canti, the songs; tu balli, you dance; i balli, the dances). This strategy allows to examine morphological alternations on identical orthographic forms. This preliminary set of words was then filtered to exclude obsolete forms and forms that are used overwhelmingly as nouns or as verbs. This scrutiny was based on judgments from 12 native Italian speakers of the same age and educational level as the participants in the MEG experiment. Words unanimously evaluated as of common use both as nouns and as verbs were retained. Words selected for the experiment ranged in length between two and four syllables. We controlled the word frequency of the experimental items as accurately as possible, since this dimension may influence brain activity in occipitotemporal regions (Kronbichler et al., 2004) and ERP amplitude (Hauk & Pulvermüller, 2004). A strict control of word frequency is very difficult in Italian, since noun morphology includes only two forms (e.g., sedia, chair; sedie, chairs), with rare exceptions (e.g., nonno, grandfather; nonni, grandfathers; nonna, grandmother, nonne, grandmothers), whereas verb morphology includes as many as 46 forms. Therefore, homonymous forms with identical frequency correspond to verb and noun lemmas of different frequencies (typically higher for verbs, which have more inflected forms); and conversely, form frequencies of homonyms with similar lemma frequencies are typically higher for nouns than for verbs. To match noun and verb homonyms as closely as possible, stimuli were selected on the basis of combined lemma and form frequency values, according to COLFIS (Bertinetto et al., 2005), which comprises 3798,275 entries. At the lemma level, verbs were more frequent than nouns; at the form level, nouns were significantly more frequent than verbs. Thus, results were unlikely to be significantly affected by this dimension. An additional matter called for attention: Italian masculine nouns take two definite articles: il (plural: i) or lo (plural: gli). Article selection is constrained by the initial sound of the noun. Lo and gli are used before nouns starting with a vowel, with z, with gn or with s þconsonant; il and i are used in all other cases. Lo and gli are also used as clitic pronouns (e.g., io lo vedo, I see him; io gli parlo, I speak to him —see Table 1), albeit much less frequently than as articles. In Italian, il is of higher frequency than lo (Bertinetto et al., 2005). Since our complex selection procedure resulted in a limited pool of usable items, it was impractical to eliminate noun homophones that take lo. To rule out that differences in frequency of usage between il and lo nouns played a relevant role on MEG activations, a frequency analysis based on COLFIS was performed on the 49 nouns that take il (mean: 21739; sd: 18452) and the 11 that take lo (mean: 19227; sd: 14589). The difference is insignificant (p ¼1). As a final step, another group of twelve native Italian speakers reviewed the 66 homophones included in the experimental list (Appendix).

2.3. Experimental paradigm Stimuli were presented on a back-projection screen, located approximately 126 cm in front of the participant inside the magnetically shielded room (MSR). The 66 homophones were included in Noun phrases (NPs) or Verb phrases (VPs), which were presented one word at a time, with the function word preceding the content word. Stimuli subtended an angle of maximally 51 at the center of the visual field. Stimulus presentation was programmed by using E-Primes 2.0 (Psychology Software Tools, Inc. Pittsburgh, PA). A photodiode attached discreetly on the projection screen detected stimulus onset and triggered the MEG acquisition system. Table 1 Center latencies (mean7 standard deviation) of the brain responses for NPs and VPs in 12 participants. Component

Mean latency (ms) Article

Pronoun

Response components FM1 FM2 FM3

100.337 18.55 160.58 725.2 255.08 7 29.68

98.58 715.54 178.75 7 29.13 258.75 742.2

CM1 CM2 CM3 CM4

Nouns 101.25 7 11.18 144.5 7 17.63 212.5 7 15.07 3417 33

Verbs 96.75 76 142.66 713.98 210.667 14.25 335.66 737.89

F¼ function word, C¼ content word.

89

To implicitly attract the subject0 s attention on the grammatical features of the homonym, stimuli consisted of pairs of phrases (NP–NP; NP–VP; VP–NP; VP–VP). As for the structure of the task, suffice it here to say it included 1024 pairs, and that there were equal numbers (n¼ 256) of: a. “identical” pairs (eg, il ballo/il ballo); b. homonym pairs that differed only in grammatical class (e.g., il ballo/io ballo); c. homonym pairs that differed only in the inflection (person for verbs, number for nouns; e.g., il ballo/i balli); d. pairs in which the first word was a homonym and the second was an orthographic neighbor (not necessarily a homonym; e.g., il ballo/il callo, the callus). In 64 pairs, equally represented in the 4 subgroups, the homonym (a noun or a verb in half the cases) was followed by a pseudoword (e.g., il ballo, followed by the pseudoword il tallo), obtained by changing a letter in a real Italian word. In all groups of stimuli, the noun was an equal number of times in the singular and in the plural form, and the verb was an equal number of times in the first or in the second singular of the present indicative. In total, the task included 1024 stimuli. Participants were instructed to silently read the stimuli on the screen while keeping their index finger on a pad, and to lift their finger when they saw a pseudoword. Responses were recorded using an optical response pad. Pairs with a pseudoword in second phrase position introduce a decision component in the silent reading task, that may interfere with cortical responses (see for example Chen, Davis, Pulvermuller, & Hauk, 2013). As a consequence, the activity elicited by the second NP/VP of each pair might be the combined result of noun/verb processing, and of decision processes. To prevent this possibility, in each participant analysis was restricted to the cortical activity associated to the first NP and VP of each pair, during which subjects were asked to read silently, without making any decision. This yielded 1024 stimuli per subject, half NPs and half VPs (n¼ 512 each). 2.4. MEG recordings Whole-head MEG recordings were obtained during the experiment at a sampling rate of 1000 Hz using a 306-channel (204 first order planar gradiometers, 102 magnetometers) VectorView MEG system (Elekta-Neuromag Ltd., Helsinki, Finland) in a magnetically shielded room (MSR). Data from both planar gradiometers and magnetometers were analyzed. Magnetometers consist of a single coil which measures the magnetic flux perpendicular to its surface. Planar gradiometers consist of a “figure-of-eight”-type arrangement of coils. The measured signal is the difference between the two loops of the “eight”, or the spatial gradient. Hardware filters were adjusted to band-pass the MEG signal in the frequency range of 0.1– 330 Hz. 2.4.1. Co-registration Prior to MEG recording, four coils were attached to the scalp of each participant, serving as head-position indicators (HPI coils). Coils were secured to the head with medical tape. The location of these coils and the head shape of each participant were recorded with a 3D digitizer (PolhemusFastrak) with respect to three anatomical landmarks: the nasion and the two-periauricular points. The four coils attached to the participant0 s head could be localized by the MEG system by providing an AC current (290, 300, 310, and 320 Hz) to the coils, that generated an oscillatory magnetic field. Thus, the headshape recorded by the 3D-digitizer could be co-registered with the MEG recording. Lacking individual structural MR images because of limited scanning resources, we used a template brain for the source estimation procedure defining the anatomical source space and the head model describing the physical relation between magnetic field and brain currents. Since the use of a template brain provides only limited localization accuracy, we refrained from detailed modeling of the cortical surface as source space constraint, and applied a two-dimensional mesh representing the surface of the brain. 2.4.2. Experimental procedure After head-shape digitization, the participant was placed in the dimly-lit MSR and comfortably seated in the MEG apparatus. The subject0 s head was placed in the helmet-shaped sensor area of the MEG. Prior to the experiment, participants were briefed in detail and completed a short training session. Subjects were instructed to fixate on the central fixation point that appeared for a pseudo-randomized interstimulus interval (5007 50 ms) and was subsequently replaced by a NP or a VP. Words from each phrase were presented, one at a time, at the center of the screen; the function word (a masculine definite article in the singular or in the plural form – il, lo, i, gli; the first or second person singular pronoun – io, tu) was shown for 275 ms and the homonym (a noun in the singular or plural form; a verb in the first or second singular person of the present indicative) for 450 ms. The timing of stimulus presentation was established on the basis of electrophysiological evidence showing that access times are considerably faster for function words than for content words, perhaps due to frequency and word predictability (Chiarello & Nuding, 1987; Segalowitz & Lane, 2000). We were aware that a longer exposure of the function word (e.g. 450 ms, like for the content word) would allow to observe the late activation linked to it, and to clearly distinguish it from content word-related activation. However, since in a phrase the homonym is assigned its grammatical class by the function word, an unnaturally long exposure of the function word might render less salient the syntactic nature of NP and VP stimuli, and ultimately the grammatical class of the content word.

90


Therefore, we opted for a time delay that was well within the range of natural reading speed, and included a shorter presentation time for determiners and pronouns than for nouns and verbs. Participants were trained to read silently the stimuli shown on the screen, and to lift their right index finger from the response pad any time they saw a pseudoword. The last word in each trial was followed by a fixation point that remained onscreen for 500 ms. The fixation point was then replaced by an ‘X’ for another 500 ms. Participants were asked to refrain from blinking during word presentation, in order to minimize eye blink artifacts in the MEG recording. They were instructed to blink freely and to provide their response when the ‘X’ symbol appeared, in order to avoid movement/muscle artifacts during stimulus presentation. The experimental session consisted of 1024 trials, and lasted approximately 56 min. It was divided into eight runs (136 trials each), each lasting approximately 7 min. Participants were allowed to rest between blocks. Special care was taken to reposition the subject0 s head after breaks such that it was as closely as possible at the same position within the MEG helmet across the whole experiment.

3. Data analysis Following acquisition, raw MEG data were analyzed using the Brain Electric Source Analysis (BESA) software package (MEGIS Software GmbH, Gräfelfing, Germany). Recorded data were offline high-pass filtered at 1.6 Hz and low-pass filtered at 60 Hz. A high-pass filter of relatively high frequency was chosen in order to reduce any possible carryover of low-frequency activity from the function word to the content word. A 50 Hz notch filter was also applied to the data. As a first step, MEG data were segmented with respect to the presentation of the first word. Epochs started 100 ms prior to first word onset and lasted until 725 ms after stimulus onset (275 ms for the functional word and 450 ms for the homonym). Epoched data were visually inspected for artifacts. Epochs contaminated by eye or head movements and/or muscle-related artifacts were discarded from further analyses. Artifact-free epoched data were then baseline-corrected by subtracting the mean activity in the 100 ms prestimulus interval for each channel, and averaged across trials for each subject and each sensor providing the evoked magnetic fields. In order to get a rough picture of the time course of activation, the global field power (GFP), i.e., the root mean square of the magnetic activity power across sensors, was computed for each time point. GFP is based on sensor level data. It corresponds to the standard deviation of the amount of activity across the scalp at each time point and reflects the power from all recording electrodes simultaneously (Skrandles, 1990). It was calculated for each participant, for each condition, and for magnetometers and gradiometers separately. Finally, evoked magnetic fields and GFP were averaged across subjects separately for the conditions under investigation. Since the experimental task included 1024 stimuli with a homonym in the first phrase, 512 responses to homonymous nouns and 512 responses to homonymous verbs were available from each of the 12 participants, for a total of 6144 responses to each stimulus type. After eliminating responses contaminated by artifacts, a minimum of 5500 artifactfree responses was used to calculate the averaged waveforms. Computation of GFP yielded a number of components that are described in Section 4. Repeated measures analysis of variance (ANOVA) was performed on GFP values to reveal differences between verbs and nouns at different time latencies. The threshold for significance was set at p o.05 (corrected for multiple comparisons; False Discovery Rate (FDR) (Benjamini & Hochberg, 1995)).

the absence of explicit a priori hypotheses, a minimum norm approach is recommended (Hauk, 2004). The Minimum Norm Estimate (MNE) models cortical activity by placing dipoles at the nodes of a grid representing either the whole brain or the cortical sheet. MNE allows to identify sources evenly distributed across the brain surface (Hämäläinen & Ilmoniemi, 1984). It does not require explicit a priori assumptions about the nature or number of source currents (Hämäläinen & Ilmoniemi, 1994; Hauk, 2004), and has been shown to depict the structure of the primary current distribution with great accuracy (Hämäläinen & Ilmoniemi, 1994; Uutela, Hämäläinen, & Somersalo, 1999). Lacking any a priori information about the involved neural generators, MNE appeared to be appropriate to identify differences between the cortical processing of nouns and verbs. MNEs were calculated for each participant individually and for all time points, by using the Minimum L2-Norm algorithm. The minimum norm images were calculated in BESA (version 5.2.1) (MEGIS Software GmbH). For each time point, up to 30 simultaneously active current sources were allowed at 1231 possible loci placed on an ellipsoid approximating an averaged cortical surface. Three orthogonal current components (in x-, y-, and z-dimensions) were calculated at each location and their root-mean-square (RMS) yielded the source strength at a given locus. Source solutions were projected on a triangularized surface of the averaged brain. The noise covariance matrix was calculated by using the individual epoch0 s baseline data (from 100 to 0 ms) from all the available trials for each subject0 s data. For regularization of the term LVLT, the noise covariance matrix, V, was calculated by using the individual epoch0 s baseline data (from 100 to 0 ms) from all the available trials for each subject0 s data. The Tikhonov regularization was used with λ¼ 0.01 (Tikhonov & Arsenin, 1977). The minimum norm images were depth-weighted as well as spatio-temporally weighted, using a signal subspace correlation measure introduced by Mosher and Leahy (1998). Dipoles placed at the nodes of the mesh were allowed to take any orientation tangential to the appropriate sphere provided by the multisphere head model realized in BESA. Statistical analysis was performed on the source level by using the cluster analysis proposed by Maris and Oostenveld (2007). Cluster analysis was used in order to avoid the problem of multiple testing that occurs if the topography of different experimental conditions is compared separately at all vertices of the cortical surface. As a first step, MNEs for NPs and VPs at each time point and each vertex were compared using paired t-tests. Only clusters consisting of at least ten spatially contiguous vertices showing a significant difference between NPs and VPs were considered for the subsequent step. Clusters sharing fewer vertices were discarded from further analysis. In the next step, contiguous clusters that persisted for at least 12 ms (12 samples) were selected. Following this procedure, each cluster contained identical vertices in both conditions. In a final step, for each cluster activity across vertices was calculated for each condition. The statistical significance of differences in cluster activations between NPs and VPs was tested using a permutation approach. The t-score obtained for the comparison between NPs and VPs was compared to a distribution of t-scores obtained by randomly assigning individual subjects0 data to one of the two conditions. As significance criterion a value of p o0.05 was chosen after p values had been Bonferroni corrected for the number of clusters.

4. Results 3.1. Minimum norm estimates (MNE)-source localization 4.1. Event-related field (ERF) analysis In neuromagnetic source estimation, neuronal brain activity is modeled by current dipoles. Depending on whether a limited number of focal sources is expected, a model composed of an a priori decided numbers of dipoles whose positions, orientations and strengths are fitted to the measured brain data might be appropriate. However, in

Brain responses induced by noun and verb processing were identified for latencies between 0 and 725 ms. Three components of event-related magnetic fields were identified for the function word (FM1, FM2, FM3), and four for the content word (CM1, CM2,


91

Fig. 1. Evoked magnetic fields for NPs (left) and VPs (right) for the magnetometers in fT. In all figures the time course of NP and VP presentation is shown in the same way. Function word and content word onsets are marked by a dotted line; the first interval (0 to 275 ms) corresponds to the presentation of the function word; the second (0 to 450 ms) to the presentation of the content word.

CM3, CM4).2 Fig. 1 represents the magnetic fields of these components for NPs and VPs. The peak latencies of the abovementioned components for noun and verb phrases are shown in Table 1. Identifiable noun and verb processing began at approximately 70 ms. Group analysis failed to reveal statistically significant differences in latency between the two grammatical classes. A fast visual response to function word onset was detected as early as 50 ms in three out of 12 subjects. Group differences in latency did not reach significance, and were not analyzed further. 4.2. GFP Fig. 2 illustrates the results of the statistical analysis for GFP, averaged across participants for NPs and VPs, for planar gradiometers (upper panel) and magnetometers (lower panel). Comparisons failed to reveal significant latency differences at component peaks. Statistically significant amplitude differences occurred at time intervals that either preceded or followed – and in some cases, partially coincided with – the mean latencies of lexical components. Gradiometer measurements showed statistically significant differences (p o.05; FDR corrected) at time intervals between 70–102 ms, 145–164 ms, and 200–237 ms after function word onset, and at 33–47 ms, 76–98 ms, 305–320 ms and 398–403 ms after content word onset. Significant differences are shown as grey stripes in Fig. 2. 4.3. Source level analysis Cluster analysis was employed to detect cortical areas in which the MNE-estimated activity was different for NPs and VPs. Fig. 3 represents the ( 410-vertex) clusters for which statistical significant differences (p o0.05; Bonferroni corrected) were observed for 4 12 consecutive time slices. Top and middle rows display the averaged MNE values of the NPs and VPs activity across the consecutive time slices for each of the 4 12-slice time windows, respectively; the bottom row shows the 4 10-vertex clusters showing significantly different activation to NPs as opposed to VPs. 2 The lack of a FM4 component reflects the fact that in our experimental paradigm the content word appeared 275 ms after the onset of the function word— that is, before the beginning of “cognitive” activity for the function word.

4.3.1. Function word (Fig. 3, green background) MNE cerebral topography showed bilateral occipital activation for articles and pronouns (Fig. 3, top and middle row). The two function words activate essentially the same structures, with the exception of increased activation for pronouns but not determiners at right parietal sites in the 88–108 ms time window and at prefrontal locations at 232–251 ms (p o0.05; FDR corrected). 4.3.2. Content word (Fig. 3, pink background) While determiners and pronouns activated roughly the same cortical regions and to a similar extent, verbs elicited substantially stronger cortical activations than nouns, especially at relatively late processing stages (Fig. 3, compare top and middle rows). On inspection, verbs yield more extensive and intense activation than nouns in several frontal, parietal and temporal areas, and nouns seem to activate the temporal pole more than verbs. Statistical comparisons (Fig. 3, bottom row) yielded significantly greater activity in response to verbs than to nouns at six time windows. The earliest activation (115–136 ms) occurred in right posterior parietal areas. In the 195–212 ms time window, a significant cluster was observed in a midline (possibly rightlateralized) superior site, encompassing centro-parietal regions. Significantly greater activation for verbs was also demonstrated in two later time windows (280–319 and 380–409 ms). Two clusters, almost overlapping in time, were documented 280–319 ms after content word onset. The first (280–300 ms) showed midline activity in central areas, similar to the earlier (195–212 ms) window, the second (297–319 ms) occurred in left inferior prefrontal regions. Two additional, almost simultaneous clusters were observed 380–409 ms post-onset of the content word. The earlier cluster (380–397 ms) was located in left posterior parietal areas, and the later cluster (393–409 ms) in left prefrontal regions, largely overlapping the 297–319 ms cluster. Areas of significantly greater activity for nouns were not observed.

5. Discussion Participants were asked to silently read homonymous nouns and verbs in a minimal syntactic context. The goal of the project was to establish if and to which extent nouns and verbs are

92


processed differently in the brain. The time course and the sources of brain activity in response to noun and verb stimuli were analysed. We used whole-head MEG to assess the spatiotemporal profile of evoked responses recorded from human healthy adults silently reading noun/verb Italian homonyms. The GFP analysis, that provides a rough measure of the temporal course of brain activation and largely ignores topographical details, revealed the well-known component structure reported for word processing for both NPs and VPs (Tarkiainen, Helenius, Hansen, Cornelissen, & Salmelin, 1999), but hardly any difference between the two. We will briefly discuss GFP results to relate our findings to the literature, but will focus mostly on the spatio-temporally refined cluster analysis. MNE analysis results showed significantly stronger activation for verbs than for nouns, at several time windows and in several brain regions. Such spatiotemporal differences have implications for an understanding of the mechanisms underlying noun and verb processing. 5.1. Response components

Fig. 2. GFP for NPs and VPs for gradiometers (upper panel) and magnetometers (lower panel). Statistically significant differences (po 0.05) between NPs and VPs occur at the time windows indicated by light-grey strips. Time windows where significant differences occurred for both sensor types are shown by dark-grey strips.

Response components of the sensor level analysis were similar for NPs and VPs. Both function and content words exhibited the early component (FM1 and CM1, respectively) referred to in the literature as M100. This response presumably reflects low-level processing, common to all visual stimuli (Tarkiainen et al., 1999). A second component (FM2 and CM2) was also observed for function and content words. Conventionally referred to as M170, it is thought to correspond to preliminary orthographic processing. It has been associated with letter string processing by Tarkiainen et al. (1999), who showed stronger left occipitotemporal activity for strings of letters than of symbols. A similar conclusion was reached by fMRI studies (e.g. Cohen et al., 2000). MEG studies also support a correlation between this component and early visual word form analysis—M170 activity was documented in pre-lexical processing of morphologically complex words (Zweig & Pylkkänen, 2008), and its intensity and timing were modulated by frequency and length (Assadollahi & Pulvermüller, 2003; Seteno, Rayner, & Posner, 1998). Components FM3 and CM3 correspond to M250, which is taken to reflect pre-lexical processing (Meng et al., 2010; Pylkkänen & Marantz, 2003; Xiang & Xiao, 2009). M350 was observed only for content words (CM4)—there could not be an FM4 component in the

Fig. 3. MNE estimates for the time windows for which significant differences between NPs and VPs were revealed by cluster analysis. The top and middle row display the averaged MNE estimates of NP- and VP-related activity across the consecutive time slices for which cluster analysis revealed significant differences between NPs and VPs. The bottom row shows the extension of clusters which significantly differed between NPs and VPs. Significant clusters were defined as the spatially and temporally contiguous brain regions ( 410 vertices; 412 samples) whose spatial extent however could vary over time. To indicate the stability of the cluster, the percentage of samples with respect to overall temporal persistence of the cluster is coded by color intensity. A value of 100% indicates that the corresponding vertex belonged to the cluster throughout the cluster0 s lifetime. Green and pink backgrounds indicate time windows corresponding to the presentation of function word and of the content word, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)


present experiment, as the content word appeared on-screen before any “cognitive” component could be detected for determiners and pronouns. M350 reflects higher-level processes than the previous components. It has been associated with lexical and semantic processing (Helenius, Salmelin, Service, & Connolly, 1998; Pylkkänen, Stringfellow, & Marantz, 2002), as suggested by the fact that its latency is frequency-sensitive (Embick, Hackl, Schaeffer, Kelepir, & Marantz, 2001). In the present study, M350 latency was indistinguishable for nouns and verbs, but several spatiotemporal differences occurred 380– 409 ms after content word onset. 5.2. Spatiotemporal differences between VPs and NPs 5.2.1. Early time window Source level analyses showed strong bilateral occipital activity for function words. In addition, cluster analysis documented stronger activity to pronouns than determiners in a right parietal site (see Fig. 3, bottom row) at 88–108 ms. For content words, a cluster of significantly greater activation for verbs than nouns, similar to the previous one as for location and latency, was observed at 115–136 ms. We have no obvious explanation for these early asymmetries. In the case of function words, it is tempting to ascribe them to physical differences between determiners and pronouns. However, this account is unlikely, as the same asymmetries are observed also for content words, which are homographs—i.e., physically identical. Early parietal activity might be due to top-down effects of task-related visual attention. This possibility is suggested by ERP studies on reading (McCarthy & Nobre, 1993; Ruz & Nobre, 2008), in which it was attributed to preparatory brain activity; and by an fMRI study (Sahin, Pinker, & Halgren, 2006) that required the production of inflected nominal and verbal forms. Also this account is unsatisfactory. In these studies, parietal activation was left-lateralized (in our study it occurred in the right hemisphere) and, more importantly, was independent of grammatical category. Yet another possibility stems from the observation that in left-toright orthographies, Optimal Viewing Position (OVP) in reading is skewed to left-of-center for words, illegal pseudowords, symbol sequences, and music (O0 Regan, Lévy-Schoen, Pynte, & Brougaillière, 1984; Brysbaert & Nazir, 2005). In our experiment, right-lateralized parietal activation might signal gaze orientation, corresponding to the leftward OVP shift. The independence of OVP from the linguistic nature of the stimulus (Wong & Hsiao, 2012; Chan, Urbach, & Kutas, 2013) is consistent with the early onset of parietal activity. However, also this hypothesis fails to account for why the effect should be greater for pronouns than determiners, and for verbs than nouns. As an alternative, noun/verb differences might result from topdown effects of morphosyntactic structure, leading to early activation of working memory (see Section 5.2.3). Participants were asked to silently read NPs and VPs. The first word could be one of four determiners (il/lo/i/gli), or one of two pronouns (io/tu). This may have set expectations for specific visual features which, once detected, may have automatically primed the verbal working memory network. The hypothesis remains very speculative, but is not inconsistent with fMRI data on morphosyntactic verb processing (Shapiro et al., 2006). Related results were obtained with MEG: in the detection of word category violations, early (125 ms) sensitivity to syntactic cues in the visual cortex was observed (Dikker, Rabagliati, & Pylkkanen, 2009). 5.2.2. Intermediate time window Significantly greater activity in response to VPs than to NPs was detected 232–251 ms after pronoun onset in anterior-superior

93

frontal areas close to the midline, and 280–319 ms after verb onset in two sources, one in the central areas between frontal and parietal cortex (280–300 ms), the other in left inferior prefrontal cortex (297–319 ms). Earlier activations to nouns and verbs (in the P200 range) were observed by Preißl, Pulvermüller, Lutzenberger, and Birbaumer (1995) and Pulvermüller et al. (1999), but not by Barber et al. (2010). Activity in the 232–319 ms time window, roughly corresponding to the M250 component, might be attributed to pre-lexical processing (Meng et al., 2010; Pylkkänen & Marantz, 2003; Xiang & Xiao, 2009). In a recent study, however, Brennan and Pylkkänen (2012) demonstrated activity around 250 ms in inferior prefrontal, medial frontal and anterior temporal areas when content words were presented in syntactic context vs in lists. Results were taken to support the hypothesis that these structures are involved in combinatorial syntactic processes. Similar areas were activated in our experiment, medial and inferior frontal activity being significantly greater in response to VPs. 5.2.3. Late time window For the purpose of this study, the most interesting finding is that 380–409 ms post-onset of the content word, conspicuous leftsided differences emerge between nouns and verbs. Activity at this time window is taken to reflect cognitive processing (Helenius et al., 1998; Pylkkänen et al., 2002). Verbs strongly activate inferior prefrontal and (more mildly) posterior parietal and temporal areas (Fig. 3, two rightmost columns). On inspection, some vertices in left anterior temporal areas show greater activity to nouns (Fig. 3, rightmost images), but the difference falls short of significance, perhaps due to the fact that statistical analyses were conducted very conservatively. In itself, the difference is interesting, as it converges with lesion and neuroimaging data showing the involvement of the temporal lobe in noun processing (e.g., Beauchamp & Martin, 2007; Capitani et al., 2009; Cotelli et al., 2006; Damasio, Grabowski, Tranel, Hichwa, & Damasio, 1996; Damasio & Tranel, 1993; Daniele et al., 1994; Martin & Chao, 2001; Martin, Ungerleider, & Haxby, 2000; Papagno, Capasso, & Miceli, 2009). However, since it falls short of significance, it will not be discussed further. Clusters significantly more active for verbs are observed in left posterior parietal (380–397 ms) and inferior prefrontal areas (393–409 ms) (Fig. 3, bottom row, two rightmost images). 5.2.3.1. Left frontal activity. Intense left inferior prefrontal activity in response to verbs is consistent with lesion studies showing a strong association between left prefrontal damage and impaired verb processing, at the conceptual and/or lexical level (Bak et al., 2001; Damasio & Tranel, 1993; Damasio et al., 2001; Daniele et al., 1994; Kemmerer, Rudrauf, Manzel, & Tranel, 2012; Tranel, Adolphs, Damasio, & Damasio, 2001; Tranel, Kemmerer, Adolphs, Damasio, & Damasio, 2003; Tranel, Martin, Damasio, Grabowski, & Hichwa, 2005), or at the morphosyntactic level (Shapiro, Shelton, & Caramazza, 2000; Laiacona & Caramazza, 2004; Tsapkini et al., 2002). It is also consistent with neuroimaging studies showing greater prefrontal activation in response to verbs, presented either as uninflected (zero-inflected) forms (e.g. Petersen et al., 1989; Raichle et al., 1994) or as inflected forms requiring morphosyntactic processing (e.g. Longe et al., 2007; Shapiro et al., 2006; Tyler, Bright, Fletcher, & Stamatakis, 2004; Wise et al., 1991). Further support comes from slowed production of inflected verbal (but not nominal) forms following rTMS of the left middle frontal gyrus (Shapiro et al., 2001; Cappelletti et al., 2008). The left prefrontal clusters observed at 297–319 ms and at 393– 409 ms have a slightly different location. With all the caution required by the less-than-optimal spatial resolution of MEG, it is tempting to speculate that the earlier, lower cluster signals

94


processing of the verb, and the later cluster, located superiorly and possibly encompassing the middle frontal gyrus, corresponds to processing and integration of inflectional verb features (Cappelletti et al., 2008). 5.2.3.2. Parietal activity. In intermediate and late time windows, verbs yielded significantly greater activation than nouns in parietal areas, almost concurrent with frontal activity. In fMRI studies, parietal activation was associated to the processing of semantic features of verbs (e.g. Hauk & Pulvermüller, 2004). This is not very likely in our case, as the asymmetry between verbs and nouns started very early. Furthermore, even if allowance is made for reduced spatial resolution of MEG, activity occurred in more posterior/inferior parietal regions than reported in fMRI studies. As an alternative, late parietal activation for verbs might signal working memory processes. The involvement of working memory during sentence processing has been demonstrated repeatedly (Baddeley, 2003; Caplan & Waters, 2000; Caplan, Michaud, & Hufford, 2013; Lewis, Vasishth, & Van Dyke, 2006; Martin, 1993). Even though the nature of its relationship with the parietal lobe is still a matter of debate (e.g., Buchsbaum & D0 Esposito, 2008), there is general consensus that working memory involves parietal structures (e.g. Borst & Anderson, 2013; Jonides, Smith, Marshuetz, & Koeppe, 1998; Oztekin, McElree, Staresina, & Davachi, 2009; Paulesu, Frith, & Frackowiak, 1993; Smith & Jonides, 1998). Analysis of the computational requirements of our VP stimuli supports this account. In Italian, inflectional features of verbs are much more complex than those of nouns—e.g., a 2nd singular pronoun can be followed by a number of verb forms, depending on tense/aspect, whereas a determiner in the singular can only be followed by a noun in the singular. In addition, pronounþverb VPs are well-formed sentences (NPs are not), and VP processing requires activating the argument structure of the verb, which must remain active for sentence interpretation. An involvement of the left parietal lobe in argument structure assignment has been shown in several lesion and fMRI studies (Meltzer-Asscher, Schuchard, den Ouden, & Thompson, 2012; Meyer, Obleser, Kiebel, & Friederici, 2012; Thompson, Bonakdarpour, & Fix, 2010; Thothathiri, Kimberg, & Schwartz, 2012). Therefore, in our experiment, late parietal activity might correspond to the recruitment of working memory processes involved in sentence processing, more specifically during tense/aspect processing and thematic role assignment. Working memory recruitment also accommodates the close temporal relationship between frontal and parietal activity, as activation in a frontoparietal network has been documented in numerous neuroimaging studies of verbal working memory (e.g. Borst & Anderson, 2013; Jonides et al., 1998; Oztekin et al., 2009; Paulesu et al., 1993; Smith & Jonides, 1998). In VP processing, frontal activity might be critical for processes underlying the integration of inflectional verb features, that must eventually interact with those involved in thematic role assignment (possibly taking place in the parietal lobe) for sentence interpretation. 5.3. The meaning of observed activations Different hypotheses have been put forth to account for noun/ verb dissociations. On one view, they are rooted in grammatical class distinctions at the lexical level (Caramazza, 1997; Shapiro & Caramazza, 2003). This view is supported by neuropsychological reports of grammatical class-by-modality dissociations in spoken vs written production of nouns vs verbs (e.g., Rapp & Caramazza, 2002; Hillis, Tuffiash, & Caramazza, 2002); of persisting grammatical class effects when semantic variables are controlled (Berndt, Haendiges, Burton, & Mitchum, 2002); and of dissociated performance on nouns/verbs and pesudonouns/pseudoverbs in morphosyntactic context (Shapiro et al., 2000; Shapiro & Caramazza,

2003). Converging evidence comes from fMRI studies showing dissociable activation to core features of verbs and nouns (Shapiro et al., 2006) and distinct activation to verbs vs nouns (collapsing across high-motion and low-motion items) but not to high-motion vs low-motion words (collapsing across nouns and verbs) (Bedny, Caramazza, Pascual-Leone, & Saxe, 2011). On this framework, noun/verb differences at the single-word level stem from lexicalgrammatical dimensions; and dissociations in syntactic context result from damage to distinct morphosyntactic processes linked to lexical-grammatical properties (Shapiro et al., 2000).3 An alternative proposal is that grammatical class is not an organizing principle of the language system, and that noun/verb distinctions rely entirely on semantic dimensions (Bird, Howard, & Franklin, 2000; Vigliocco et al., 2011). Consistent with this possibility, in picture-word interference tasks healthy subjects showed grammatical class effects at intervals typical of semantic processing (Pechmann, Garrett, & Zebst, 2004). Additional evidence comes from fMRI data (Siri et al., 2008) showing activation in the inferior frontal gyrus regardless of whether subjects named the same pictured event with an action noun (corsa, the run), an inflected action verb (corre, s/he runs) or an action verb in the infinitive (correre, to run). In a single-word ERP study, Barber et al. (2010) presented nouns and verbs referring to motor and sensory events. Both grammatical class (noun/verb) and semantic features (motion/sensation) yielded significant N400 effects, indistinguishable for latency, duration, and scalp distribution. This pattern of results was interpreted as evidence for a single underlying process, driven by conceptual properties. Proponents of this view argue that, when conceptual dimensions (e.g., the event nature of nouns and verbs) are fully matched, there is no evidence that grammatical class constrains language/brain relationships. On this account, verbs and nouns as isolated words are processed by the same substrate. Dissociations at the single-word level result from anatomically distinct representations of conceptual properties, and dissociations in morphosyntactic context reflect greater processing demands for verbs in a shared neural substrate (for review, see Vigliocco et al., 2011). Previous ERP and MEG studies showed distinct brain waveforms in response to nouns and verbs (e.g. Federmeier et al., 2000; Fiebach et al., 2002; Koenig & Lehmann, 1996; Pulvermüller et al., 1996, 1999; Xiang & Xiao, 2009), providing some support to the notion that the two word types are processed in at least partially distinct neural substrates. Spatial information from MEG studies shows distinct or stronger activity in the inferior frontal gyrus for verbs and in the temporal lobe for nouns (Pulvermüller et al., 1999; Xiang & Xiao, 2009), consistent with fMRI (Shapiro et al., 2006; Tyler et al., 2004; Longe et al., 2007), and rTMS studies (Shapiro et al., 2001; Cappelletti et al., 2008). Available MEG results, however, do not establish whether differences result from intrinsic grammatical properties, conceptual distinctions, or interactions between morphosyntactic properties and processing resources. Our results allow to discuss these issues. We wished to establish whether NPs and VPs are treated differently in the brain, and focused on processing homonyms in minimal syntactic context. Selection of homonyms allowed to eliminate nuisance factors that may affect MEG activity during written word processing and/ or phonological retrieval (Assadollahi & Pulvermüller, 2003; Seteno et al., 1998), but greatly limited the number of usable

3 This view acknowledges that the semantic representations of nouns and verbs differ under critical respects, and that in some cases such differences could justify noun/verb dissociations (Berndt, Haendiges, Burton, & Mitchum, 2002). It argues, however, that grammatical class is an organizing principle of language in the brain and that, at least in some cases, a lexical–grammatical distinction provides the best account of observed dissociations.


stimuli. This prevented the possibility to prepare lists that would enable to contrast grammatical class and semantic influences on NP/VP processing. With these limitations in mind, some considerations are possible. Target noun/verb homonyms had largely overlapping semantic features. Over 90% of the nouns referred to events, and were very close to verbs in meaning (e.g., ballo 4the dance/I dance; bacio 4the kiss/I kiss). On these premises, the semantic account would predict almost indistinguishable activations. By contrast, verbs but not nouns yielded prefrontal (195–212 ms; 297–319 ms; 393–409 ms) and parietal activity (380–397 ms)—see Fig. 3. These differences are unlikely to result from semantic variables, and invite the conclusion that at least partly separable neural substrates are involved in processing grammatical class information and the corresponding morphosyntactic operations. Consistent with this view, an fMRI study of object nouns, event nouns and verbs (Garbin, Collina, & Tabossi, 2012) showed a pattern of both separate and shared areas of BOLD activation. This conclusion is mitigated by other observations. Even though verbs yielded significantly greater activity at six time windows, many differences were quantitative more than qualitative—the same vertices were activated by verbs and nouns, only to a greater extent by the former. This might reflect more intense (and/or more extensive) activation in a shared neural substrate, possibly due to greater processing demands of verb morphosyntax, as proposed by the semantic hypothesis (Vigliocco et al., 2011).4 Interestingly, many observations consistent with this hypothesis were obtained from Italian, whose verb morphology is substantially more complex than noun morphology. In English, that has comparably simple inflectional systems for nouns and verbs, double dissociations were observed both in aphasics (Shapiro & Caramazza, 2003; Shapiro et al., 2000) and healthy controls (Shapiro et al., 2006). Overall, results are consistent with the lexical account. However, they do not establish the extent in which conceptual, lexical and morphosyntactic verb properties each contribute to MEG activity, and therefore do rule out the semantic hypothesis.

6. Conclusions When tested with homonyms, the neural mechanisms involved in noun/verb processing show subtle but clear distinctions. In the present MEG study, sensor level analyses demonstrated only very short-lived differences. However, source level analyses revealed significantly stronger left inferior prefrontal and posterior parietal activity to verbs at 280–319 ms and mostly at 380–409 ms. These differences, albeit substantial, may be too subtle to be picked up by fMRI, during which spatiotemporal distinctions could be blurred by overlapping effects of cognitive processes that are temporally too close to be discriminated by techniques relying on metabolic changes. The present study supports the view that processing NPs and VPs involves partially distinct neural substrates, but does not rule out that NPs and VPs might recruit different amounts of neural tissue in shared networks. It confirms the critical role of left inferior prefrontal cortex in VP processing, and suggests that verb processing, even in a minimal syntactic context, recruits a frontoparietal working memory network. The neural substrate involved in MEG activity for verbs is fully consistent with lesion data in subjects with impaired verb processing. Converging results from independent sources of evidence show that MEG studies can contribute to the ongoing discussion on noun/verb distinctions. 4 For apparently shared areas of activation, it is also possible that a macroscopic brain region houses interwoven cell assemblies that belong to distinct neural networks, and are recruited only by one type of stimulus.

95

Acknowledgments We gratefully thank Dr. Elena Betta and Elisa Leonardelli for their technical support with the experimental setup and the recordings. We also thank two anonymous reviewers for their helpful suggestions and criticisms. This work has been realized with support from the Provincia Autonoma di Trento, the Fondazione Cassa di Risparmio di Trento e Rovereto, the German Ministry for Research and Technology and the Center of Excellence of the University of Tübingen.

Appendix List of the homonyms used for the experiment. The form that corresponds to the singular of the noun/1st person singular of the verb is reported. For all homonyms, the form corresponding to the plural (for the noun) and to the 2nd singular (for the verb) is obtained by substituting the letter i for the final letter o. Homonyms that take lo/gli as determiner when used in noun context abbraccio; addebito; anticipo; arredo; incasso; invito; ostacolo; spaccio; spreco; spunto; stimolo. Homonyms that take il/i as determiner when used in noun context bacio; bagno; ballo; baro; blocco; calcolo; cammino; canto; castigo; circolo; compenso; concilio; conteggio; contrasto; covo; digiuno; disturbo; dono; filo; filtro; fischio; frammento; freno; fumo; grido; guadagno; martello; massaggio; noleggio; parcheggio; presidio; raduno; reclamo; regalo; regno; respiro; restauro; rialzo; ricambio; ricavo; rifugio; rigetto; rilancio; rimbalzo; rinnovo; risparmio; risveglio; ritocco; rotolo; saccheggio; salto; solco; sospiro; suono; taglio

References Assadollahi, R., & Pulvermüller, F. (2003). Early influences of word length and frequency: A group study using MEG. Neuroreport, 14(8), 1183–1187. Baddeley, A. (2003). Working memory: Looking back and looking forward. Nature Reviews, Neuroscience, 4, 829–839. Bak, T. H., O0 Donovan, D. G., Xuereb, J. H., Boniface, S., & Hodges, J. R. (2001). Selective impairment of verb processing associated with pathological changes in Brodmann areas 44 and 45 in the Motor Neuron Disease-Dementia-Aphasia syndrome. Brain, 124(1), 103–120. Barber, H., Kousta, S.-T., Otten, L. J., & Vigliocco, G. (2010). Event-related potentials to event-related words: Grammatical class and semantic attributes in the representation of knowledge. Brain Research, 1332, 65–74. Baroni, M., Bernardini, S., Ferraresi, A., & Zanchetta, E. (2009). The WaCky wide web: A collection of very large linguistically processed web-crawled corpora. Journal of Language Resources and Evaluation, 43(3), 209–226. Beauchamp, M. S., & Martin, A. (2007). Grounding object concepts in perception and action: Evidence from fMRI studies of tools. Cortex, 43, 461–468. Bedny, M., Caramazza, A., Grossman, E., Pascual-Leone, A., & Saxe, R. (2008). Concepts are more than percepts: The case of action verbs. The Journal of Neuroscience, 28(44), 11347–11353. Bedny, M., Caramazza, A., Pascual-Leone, A., & Saxe, R. (2011). Typical neural representations of action verbs develop without vision. Cerebral Cortex, 22(2), 286–293. Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, 57, 289–300. Berlingeri, M., Crepaldi, D., Roberti, R., Scialfa, G., Luzzatti, C., & Paulesu, E. (2008). Nouns and verbs in the brain: Grammatical class and task specific effects as revealed by fMRI. Cognitive Neuropsychology, 25, 528–558. Berndt, R. S., Haendiges, A. ,N., Burton, M. W., & Mitchum, C. C. (2002). Grammatical class and imageability in aphasic word production: Their effects are independent. Journal of Neurolinguistics, 15, 353–371. Bertinetto, P. M., Burani, C., Laudanna, A., Marconi, L., Ratti, D., Rolando, C., et al. (2005). Corpus e Lessico di Frequenza dell0 Italiano Scritto (CoLFIS) [Corpus and Frequency Lexicon of written Italian (CoLFIS)]. Available at www.istc.cnr.it/ material/database/colfis/. Bird, H., Howard, D., & Franklin, S. (2000). Why is a verb like an inanimate object? Grammatical category and semantic category deficits. Brain and Language, 72, 246–309. Borst, J. P., & Anderson, J. R. (2013). Using model-based functional MRI to locate working memory updates and declarative memory retrievals in the fronto-

96


parietal network. Proceedings of the National Academy of Sciences (USA), 110, 1628–1633. Brennan, J., & Pylkkänen, L. (2012). The time-course and spatial distribution of brain activity associated with sentence processing. NeuroImage, 60, 1139–1148. Brysbaert, M., & Nazir, T. (2005). Visual constraints in written word recognition: Evidence from the optimal viewing position effect. Journal of Research in Reading, 28, 216–228. Buchsbaum, B. R., & D0 Esposito, M. (2008). The search for the phonological store: From loop to convolution. Journal of Cognitive Neuroscience, 20, 762–778. Capitani, E., Laiacona, M., Pagani, R., Capasso, R., Zampetti, P., & Miceli, G. (2009). Posterior cerebral artery infarcts and semantic category dissociations. A study of 28 patients. Brain, 132, 965–981. Caplan, D., Michaud, J., & Hufford, R. (2013). Dissociations and associations of performance in syntactic comprehension in aphasia and their implications for the nature of aphasic deficits. Brain and Language,, 127(1), 21–33. Caplan, D., & Waters, G. S. (2000). Verbal working memory and sentence comprehension. Behavioral and Brain Sciences, 22(1), 77–94. Cappa, S. F., Sandrini, M., Rossini, P. M., Sosta, K., & Miniussi, C. (2002). The role of the left frontal lobe in action naming: rTMS evidence. Neurology, 59, 720–723. Cappelletti, M., Fregni, F., Shapiro, K., Pascual-Leone, A., & Caramazza, A. (2008). Processing nouns and verbs in the left frontal cortex: A transcranial magnetic stimulation study. Journal of Cognitive Neuroscience, 20, 707–720. Caramazza, A. (1997). How many levels of processing are there in lexical access? Cognitive Neuropsychology, 14(177), 208. Caramazza, A., & Hillis, A. (1991). Lexical organization of nouns and verbs in the brain. Nature, 349, 788–790. Chan W. S., Urbach, T. P., & Kutas, M.. (2013). The hemispheric differences on the optimal viewing position asymmetry. In: Poster presented at the neurobiology of language conference, San Diego; November, 2013. Chen, Y., Davis, M. H., Pulvermüller, F., & Hauk, O. (2013). Task modulation of brain responses in visual word recognition as studied using EEG/MEG and fMRI. Frontiers in Human Neuroscience, 7, 376, http://dx.doi.org/10.3389/fnhum.2013.00376. Chiarello, C., & Nuding, S. (1987). Visual field effects for processing content and function words. Neuropsychologia, 25(3), 539–548. Cohen, L., Dehaene, S., Naccache, L., Lehericy, S., Dehaene–Lambertz, G., Henaff, M. A., et al. (2000). The visual word form area: Spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients. Brain, 123, 291–307. Cotelli, M., Borroni, B., Manenti, R., Alberici, A., Calabria, M., Agosti, C., et al. (2006). Action and object naming in frontotemporal dementia, progressive supranuclear palsy, and corticobasal degeneration. Neuropsychology, 20, 558–565. Crepaldi, D., Berlingeri, M., Paulesu, E., & Luzzatti, C. G. (2011). A place for nouns and a place for verbs? A critical review of neurocognitive data on grammatical-class effects. Brain and Language, 116(1), 33–49. Damasio, A., & Tranel, D. (1993). Nouns and verbs are retrieved with differentially distributed neural systems. Proceedings of the National Academy of Sciences of the United States of America, 90, 4957–4960. Damasio, H., Grabowski, T. J., Tranel, D., Hichwa, R. D., & Damasio, A. R. (1996). A neural basis for lexical retrieval. Nature, 380, 499–505. Damasio, H., Grabowski, T. J., Tranel, D., Ponto, L. L. B., Hichwa, R. D., & Damasio, A. R. (2001). Neural correlates of naming actions and of naming spatial relations. NeuroImage, 13(6), 1053–1064. Daniele, A., Giustolisi, L., Silveri, M. C., Colosimo, C., & Gainotti, G. (1994). Evidence for a possible neuroanatomical basis for lexical processing of nouns and verbs. Neuropsychologia, 32, 1325–1341. Davis, M. H., Meunier, F., & Marslen-Wilson, W. D. (2004). Neural responses to morphological, syntactic, and semantic properties of single words: An fMRI study. Brain and Language, 89, 439–449. Dehaene, S. (1995). Electrophysiological evidence for category specific word processing in the normal human brain. NeuroReport, 6, 2153–2157. Dikker, S., Rabagliati, H., & Pylkkänen, L. (2009). Sensitivity to syntax in visual cortex. Cognition, 110(3), 293–321. Embick, D., Hackl, M., Schaeffer, J., Kelepir, M., & Marantz, M. (2001). A magnetoencephalographic component whose latency indexes lexical frequency. Cognitive Brain Research, 10, 345–348. Federmeier, K. D., Segal, J. B., Lombrozo, T., & Kutas, M. (2000). Brain responses to nouns, verbs and class-ambiguous words in context. Brain, 123(12), 2552–2566. Fiebach, C. J., Maess, B., & Friederici, A. D.. (2002) Neuromagnetic evidence that differences in verb and noun processing are modulated by the presence of a syntactic context. In: Haueisen, J., Nowak, H., Gießler, F., Huonker, R. (Eds.), In: Proceedings of 13th international conference on biomagnetism. Jena, Germany, 339–341. Finocchiaro, C., Basso, G., Giovenzana, A., & Caramazza, A. (2010). Morphological complexity reveals verb-specific prefrontal engagement. Journal of Neurolinguistics, 23, 553–563. Garbin, G., Collina, S., & Tabossi, P. (2012). Argument structure and morphological factors in noun and verb processing: An fMRI study. PLoS One, 7(9), e45091. http://dx.doi.org/10.1371/journal.pone.0045091. Gomes, H., Ritter, W., Tartter, V. C., Vaughan, H. G., Jr, & Rosen, J. J. (1997). Lexical processing of visually and auditorily presented nouns and verbs: Evidence from reaction time and N400 priming data. Cognitive Brain Research, 6(2), 121–134. Hämäläinen, M. S., & Ilmoniemi, R. J.. (1984). Interpreting measured magnetic fields of the brain: Minimum norm estimates of current distributions. Helsinki University of Technology, Technical Report TKK-F-A559.

Hämäläinen, M. S., & Ilmoniemi, R. J. (1994). Interpreting magnetic fields of the brain: Minimum norm estimates. Medical and Biological Engineering and Computing, 32, 35–42. Hillis, A. E., Tuffiash, E., & Caramazza, A. (2002). Modality-specific deterioration in naming verbs in nonfluent primary progressive aphasia. Journal of Cognitive Neuroscience, 14(7), 1099–1108. Hauk, O. (2004). Keep it simple: A case for using classical minimum norm estimation in the analysis of EEG and MEG data. NeuroImage, 21, 1612–1621. Hauk, O., & Pulvermüller, F. (2004). Effects of word length and frequency on the human event-related potential. Clinical Neurophysiology, 115, 1090–1103. Helenius, P., Salmelin, R., Service, E., & Connolly, J. F. (1998). Distinct time courses of word and context comprehension in the left temporal cortex. Brain, 121, 1133–1142. Jonides, J., Smith, E. E., Marshuetz, C., & Koeppe, R. A. (1998). Inhibition in verbal working memory revealed by brain activation. Proceedings of the National Academy of Sciences, 95, 8410–8413. Kemmerer, D., Rudrauf, D., Manzel, K., & Tranel, D. (2012). Behavioral patterns and lesion sites associated with impaired processing of lexical and conceptual knowledge of actions. Cortex, 48, 826–848. Khader, P. H., & Rösler, F. (2004). EEG power and coherence analysis of visually presented nouns and verbs reveals left frontal processing differences. Neuroscience Letters, 354(2), 111–114. Koenig, T., & Lehmann, D. (1996). Microstates in language-related brain potential maps show noun-verb differences. Brain and Language, 53, 169–182. Kronbichler, M., Hutzler, F., Wimmer, H., Mair, A., Staffen, W., & Ladurner, G. (2004). The visual word form area and the frequency with which encountered: Evidence from a parametric fMRI study. NeuroImage, 21, 946–953. Laiacona, M., & Caramazza, A. (2004). The noun/verb dissociation in language production: Varieties of causes. Cognitive Neuropsychology, 21(2), 103–123. Lewis, R. L., Vasishth, S., & Van Dyke, J. A. (2006). Computational principles of working memory in sentence comprehension. Trends in Cognitive Sciences, 10 (10), 447–454. Liljeström, M., Hultén, A., Parkkonen, L., & Salmelin, R. (2009). Comparing MEG and fMRI views to naming actions and objects. Human Brain Mapping, 30, 1845–1856. Longe, O., Randall, B., Stamatakis, E. A., & Tyler, L. K. (2007). Grammatical categories in the brain: The role of morphological structure. Cerebral Cortex, 17(8), 1812–1820. Maris, E., & Oostenveld, R. (2007). Nonparametric statistical testing of EEG- and MEG-data. Journal of Neuroscience Methods, 164, 177–190. Martin, R. (1993). Short-term memory and sentence processing: Evidence from neuropsychology. Memory & Cognition, 21(2), 176–183. Martin, A., Ungerleider, L. G., & Haxby, J. V. (2000). In: M. S. Gazzaniga (Ed.), Category-specificity and the brain: The sensory-motor model of semantic representations of objects. The new cognitive neurosciences (2nd ed.). Cambridge: MIT Press. Martin, A., & Chao, L. L. (2001). Semantic memory and the brain: Structure and processes. Current Opinion in Neurobiology, 11, 194–201. McCarthy, G., & Nobre, A. C. (1993). Modulation of semantic processing by spatial selective attention. Electroencephalography and Clinical Neurophysiology, 88, 210–229. Meltzer-Asscher, A., Schuchard, J., den Ouden, D. B., & Thompson, C. K. (2012). The neural substrates of complex argument structure representations: Processing “alternating transitivity” verbs. Language and Cognitive Processes, 28(8), 1154–1168. Meng, L., Xiang, J., Rose, D., Kotecha, R., Vannest, J., Byars A., et al. (2010) Time course and neural network for comparing written and spoken words: A MEG and DTI study. In: Proceedings of 17th international conference on biomagnetism. Advances in biomagnetism. Dubrovnik, Croatia. Meyer, L., Obleser, J., Kiebel, S. J., & Friederici, A. D. (2012). Spatiotemporal dynamics of argument retrieval and reordering: An FMRI and EEG study on sentence processing. Frontiers in Psychology, 3, 523, http://dx.doi.org/10.3389/ fpsyg.2012.00523. Miceli, G., Silveri, M., Villa, G., & Caramazza, A. (1984). On the basis for the agrammatic0 s difficulty in producing main verbs. Cortex, 20, 207–220. Mosher, J. C., & Leahy, R. M. (1998). Recursive MUSIC: A framework for EEG and MEG source localization. IEEE Transactions on Biomedical Engineering, 45(11), 1342–1354. 0 O Regan, J. K., Lévy-Schoen, A., Pynte, J., & Brugaillère, B. (1984). Convenient fixation location within isolated words of different length and structure. Journal of Experimental Psychology: Human Perception and Performance, 10(2), 250–257. Oztekin, I., McElree, B., Staresina, B. P., & Davachi, L. (2009). Working memory retrieval: Contributions of the left prefrontal cortex, the left posterior parietal cortex, and the hippocampus. Journal of Cognitive Neuroscience, 21, 581–593. Papadelis, C., Poghosyan, V., Fenwick, P. B., & Ioannides, A. A. (2009). MEG0 s ability to localise accurately weak transient neural sources. Clinical Neurophysiology, 120(11), 1958–1970. Papagno, C., Capasso, R., & Miceli, G. (2009). Reversed concreteness effect for nouns in a subject with semantic dementia. Neuropsychologia0028-393247(4), 1138–1148. Paulesu, E., Frith, C. D., & Frackowiak, R. S. J (1993). The neural correlates of the verbal component of working memory. Nature, 362, 342–344. Pechmann, T., Garrett, M., & Zebst, D. (2004). The time course of receovery for grammatical category information during lexical processing for syntactic construction. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30(3), 723–728.


Perani, D., Cappa, S., Schnur, T., Tettamanti, M., Collina, S., Rosa, M., et al. (1999). The neural correlates of verb and noun processing: A PET study. Brain, 122, 2337–2344. Petersen, S., Fox, P., Posner, M., Mintum, M., & Raichle, M. (1989). Positron emission tomographic studies of the processing of single words. Journal of Cognitive Neuroscience, 1, 153–170. Preißl, H., Pulvermüller, F., Lutzenberger, W., & Birbaumer, N. (1995). Evoked potentials distinguish btween nouns and verbs. Neuroscience Letters, 197(1), 81–83. Pulvermüller, F., Preißl, H., Lutzenberger, W., & Birbaumer, N. (1996). Brain rhythms of language: Nouns versus verbs. European Journal of Neuroscience, 8, 937–941. Pulvermüller, F., Preißl, H., & Lutzenberger, W. (1999). Nouns and verbs in the intact brain: Evidence from event-related potentials and high-frequency cortical responses. Cerebral Cortex, 9, 497–506. Pulvermüller, F., Cook, C., & Hauk, O. (2012). Inflection in action: Semantic motor system activation to noun- and verb-containing phrases is modulated by the presence of overt grammatical markers. NeuroImage, 60(2), 1367–1379. Pylkkänen, L., Stringfellow, A., & Marantz, A. (2002). Neuromagnetic evidence for the timing of lexical activation: An MEG component sensitive to phonotactic probability but not to neighborhood density. Brain and Language, 81, 666–678. Pylkkänen, L., & Marantz, A. (2003). Tracking the time course of word recognition with MEG. Trends in Cognitive Sciences, 7, 187–189. Raichle, M., Fiez, J., Videen, T., MacLeod, A. -M., Pardo, J., Fox, P., et al. (1994). Practice-related changes in human brain functional anatomy during nonmotor learning. Cerebral Cortex, 4, 8–26. Ramkumar, P., Jas, M., Pannasch, S., Hari, R., & Parkkonen, L. (2013). Feature-specific information processing precedes concerted activation in human visual cortex. Journal of Neuroscience, 33(18), 7691–7699. Rapp, B., & Caramazza, A. (2002). Selective difficulties with spoken nouns and written verbs: A single case study. Journal of Neurolinguistics, 15, 373–402. Ruz, M., & Nobre, A. C. (2008). Dissociable top-down anticipatory neural states for different linguistic dimensions. Neuropsychologia, 46(4), 1151–1160. Sahin, N. T., Pinker, S., & Halgren, E. (2006). Abstract grammatical processing of nouns and verbs in Broca0 s area: Evidence from fMRI. Cortex, 42(4), 540–562. Segalowitz, S. J., & Lane, K. (2000). Lexical access of function versus content words. Brain and Language, 75, 376–389. Seteno, S. C., Rayner, K., & Posner, M. I. (1998). Establishing a time-line of word cognition: Evidence from eye movements and event-related potentials. Neuroreport, 9(10), 2195–2200. Shapiro, K., Shelton, J., & Caramazza, A. (2000). Grammatical class in lexical production and morhpological processing: Evidence from a case of fluent aphasia. Cognitive Neuropsychology, 17(8), 665–682. Shapiro, K., Pascual-Leone, A., Mottaghy, F. M., Gangitano, M., & Caramazza, A. (2001). Grammatical distinction in the left frontal cortex. Journal of Cognitive Neuroscience, 13, 713–720. Shapiro, K., & Caramazza, A. (2003). Grammatical processing of nouns and verbs in left frontal cortex? Neuropsychologia, 41, 1189–1198. Shapiro, K., Moo, L., & Caramazza, A. (2006). Cortical signatures of noun and verb production. Proceedings of the National Academy of Sciences of the United States of America, 103(5), 1644–1649. Siri, S., Tettamanti, M., Cappa, S. F., Della Rosa, P., Saccuman, C., Scifo, P., et al. (2008). The neural substrate of naming events: Effects of processing demands but not of grammatical class. Cerebral Cortex, 18, 171–177. Skrandles, W. (1990). Global field power and topographical similarity. Brain Topography, 3, 137–141. Smith, E. ,E., & Jonides, J. (1998). Neuroimaging analyses of human working memory. Proceedings of the National Academy of Science, 95, 12061–12068.

97

Sörös, P., Cornelissen, K., Laine, M., & Salmelin, R. (2003). Naming actions and objects: Cortical Nouns and verbs in the brain dynamics in healthy adults and in an anomic patient with a dissociation in action/object naming. NeuroImage, 19, 1787–1801. Tarkiainen, A., Helenius, P., Hansen, P. C., Cornelissen, P. L., & Salmelin, R. (1999). Dynamics of letter string perception in the human occipitotemporal cortex. Brain, 122(11), 2119–2132. Thompson, C. K., Bonakdarpour, B., & Fix, S. F. (2010). Neural mechanisms of verb argument structure processing in agrammatic aphasic and healthy agematched listeners. Journal of Cognitive Neuroscience, 22(9), 1993–2011. Tikhonov, A. N., & Arsenin, V. Y. (1977). Solution of Ill-posed problems. Washington: Winston & Sons0-470-99124-0. Thothathiri, M., Kimberg, D. Y., & Schwartz, M. F. (2012). The neural basis of reversible sentence comprehension: Evidence from voxel-based lesion-symptom mapping in aphasia. Journal of Cognitive Neuroscience, 24, 212–222. Tranel, D., Adolphs, R., Damasio, H., & Damasio, A. R. (2001). A neural basis for the retrieval of words for actions. Cognitive Neuropsychology, 18, 655–674. Tranel, D., Kemmerer, D., Adolphs, R., Damasio, H., & Damasio, A. R. (2003). Neural correlates of conceptual knowledge for actions. Cognitive Neuropsychology, 20, 409–432. Tranel, D., Martin, C., Damasio, H., Grabowski, T. J., & Hichwa, R. (2005). Effects of noun-verb homonymy on the neural correlates of naming concrete entities and actions. Brain and Language, 92, 288–299. Tsapkini, K., Jarema, G., & Kehayia, E. (2002). A morphological processing deficit in verbs but not in nouns: A case study in a highly inflected language. Journal of Neurolinguistics, 15, 265–288. Tyler, L. K., Russell, R., Fadili, J., & Moss, H. E. (2001). The neural representation of nouns and verbs: PET studies. Brain, 124(8), 1619–1634. Tyler, L. K., Bright, P., Fletcher, P., & Stamatakis, E. A. (2004). Neural processing of nouns and verbs: The role of inflectional morphology. Neuropsychologia, 42, 512–523. Tyler, L., Randall, B., & Stamatakis, E. (2008). Cortical differentiation for nouns and verbs depends on grammatical markers. Journal of Cognitive Neuroscience, 20, 1381–1389. Uutela, K., Hämäläinen, M., & Somersalo, E. (1999). Visualization of magnetoencephalographic data using minimum current estimates. NeuroImage, 10, 173–180. Warburton, E., Wise, R., Price, C., Weiller, C., Hadar, U., Ramsay, S., et al. (1996). Noun and verb retrieval by normal subjects: Studies with PET. Brain, 119, 159–179. Vigliocco, G., Vinson, D. ,P., Druks, J., Barber, H., & Cappa, S. F. (2011). Nouns and verbs in the brain: A review of behavioural, electrophysiological, neuropsychological and imaging studies. Neuroscience and Biobehavioral Reviews, 35, 407–426. Wise, R., Chollet, F., Hadar, U., Friston, K., Hoffner, E., & Frackowiak, R. (1991). Distribution of cortical neural networks involved in word comprehension and word retrieval. Brain, 114, 1803–1817. Wong Y. K., & Hsiao, J. H. (2012) Reading direction is sufficient to account for the optimal viewing position in reading: The case of music reading. In: CogSci 2012 proceedings, 2540-2545. Xiang, J., & Xiao, Z. (2009). Spatiotemporal and frequency signatures of noun and verb processing: A wavelet-based beamformer study. Journal of Clinical Experimental Neuropsychology, 31(6), 648–657. Yokoyama, S., Miyamoto, T., Riera, J., Kim, J., Akitsuki, Y., Iwata, K., et al. (2006). Cortical mechanisms involved in the processing of verbs: An fMRI study. Journal of Cognitive Neuroscience, 18, 1304–1313. Zweig, E., & Pylkkänen, L. (2008). A visual M170 effect of morphological complexity. Language and Cognitive Processes, 24(3), 412–439.

Electrical stimulation mapping of nouns and verbs in Broca's area.

Lexical organization of nouns and verbs in the brain.

Recognition of verbs, abstract nouns and concrete nouns from the left and right visual half-fields.

Retrieval of nouns and verbs in agrammatism and anomia.

Production of nouns and verbs in picture naming and narrative tasks by Chinese speakers with aphasia.

Lexical Retrieval of Nouns and Verbs in a Sentence Completion Task.

Embodied language in first- and second-language speakers: neural correlates of processing motor verbs.

Nouns, verbs, objects, actions, and abstractions: local fMRI activity indexes semantics, not lexical categories.

Are Nouns Learned Before Verbs? Infants Provide Insight into a Longstanding Debate.

Gavagai is as Gavagai does: learning nouns and verbs from cross-situational statistics.

How handshape type can distinguish between nouns and verbs in homesign.

Infants use known verbs to learn novel nouns: evidence from 15- and 19-month-olds.

Compositionality and the angular gyrus: A multi-voxel similarity analysis of the semantic composition of nouns and verbs.

Children's acquisition of nouns and verbs in Italian: contrasting the roles of frequency and positional salience in maternal language.

The neural basis of sublexical speech and corresponding nonspeech processing: a combined EEG-MEG study.

The fox and the cabra: an ERP analysis of reading code switched nouns and verbs in bilingual short stories.

Reply to Commentary on "Neural correlates of mirth and laughter: A direct electrical cortical stimulation study".

Maturation of auditory neural processes in autism spectrum disorder - A longitudinal MEG study.

The Neural Correlates of Shoulder Apprehension: A Functional MRI Study.

Where are the cookies? Two- and three-year-olds use number-marked verbs to anticipate upcoming nouns.

Neural dynamics of speech act comprehension: an MEG study of naming and requesting.

Pay What You Want! A Pilot Study on Neural Correlates of Voluntary Payments for Music.

Hyperactivity Disorder.

Neural correlates of gratitude.