Skip to main content
NCBI home page
Human Brain Mapping logoLink to Human Brain Mapping
. 2009 Nov 3;31(5):770–785. doi: 10.1002/hbm.20904

Cortical representation of verbs with optional complements: The theoretical contribution of fMRI

Einat Shetreet 1,✉, Naama Friedmann 2, Uri Hadar 1
PMCID: PMC6870685  PMID: 19890846

Abstract

Verbs like “eat” are special in that they can appear both with a complement (e.g., “John ate ice‐cream”) and without a complement (“John ate”). How are such verbs with optional complements represented? This fMRI study attempted to provide neurally based constraints for the linguistic theory of the representation of verbs with optional complements. One linguistic approach suggests that the representation of these verbs in the lexicon includes two complementation frames (one with and one without the complement), similarly to verbs that allow two different types of complements (e.g., discover). Another approach assumes that only one frame is represented (with a complement) and, when the complement is omitted, the relevant thematic role is saturated, either lexically or syntactically. We compared the patterns of cortical activation of verbs with optional complements to verbs that take either one or two frames and to verbs with one or two complements. These comparisons—together with prior findings regarding the cortical activation related to the number of complementation frames and the number of complements—were used to decide between the theoretical approaches. We found support for the idea that verbs with optional complements have only one frame and that a lexical operation enables complement omission. We also used fMRI in the traditional manner and identified the fusiform gyrus and the temporo–parieto–occipital junction as the regions that participate in the execution of the omission and saturation of optional complements. Hum Brain Mapp, 2010. © 2009 Wiley‐Liss, Inc.

Keywords: fusiform gyrus, lexicon, neurolinguistics, precuneus, syntax, Wernicke's area

INTRODUCTION

Imaging studies with fMRI are typically used for the localization of cognitive function, but, in some conditions, they can also inform cognitive theories by providing critical evidence in support of one of a number of hypotheses. This study is of that nature, aiming to adduce evidence that may help to decide among current theories regarding the representation of verbs with optional complements.

The knowledge of a language speaker about verbs includes, in addition to knowledge about the meaning and sound of verbs, the syntactic structure of the sentences in which each verb can appear. There are various aspects to this lexical‐syntactic knowledge, but some of the most crucial aspects concern the complements of the verb. Complements are entities participating in the event described by the verb, beyond and in addition to the subject. Some verbs, such as sneeze, do not allow any complement, and therefore a sentence like (1a) is grammatical, whereas a sentence like (1b) is not (its ungrammaticality is marked by an asterisk). Other verbs, like punish, require one complement. Thus, sentence (2a) is ungrammatical, because the complement is missing, whereas a sentence with a complement, like (2b), is grammatical. The focus of our study is a third group of verbs, for which the complement is optional. For example, the verb eat can appear with or without a complement and hence, both (3a) and (3b) are grammatical.

  • 1
    • (a)
      John sneezed.
    • (b)
      *John sneezed [the wall].
  • 2
    • (a)
      *John punished.
    • (b)
      John punished [the boy].
  • 3
    • (a)
      John ate.
    • (b)
      John ate [the corn].
  • 4
    • (a)
      John discovered [the truth].
    • (b)
      John discovered [that the story was true].

Some verbs can be complemented by one of several types of complements. For example, the verb discover can be complemented by either a noun phrase (example 4a) or a sentence (example 4b). The number of possible options of complementation and the types of the complements are specified by the complementation frames of the verb [Chomsky, 1965; Grimshaw, 1979; Van Valin, 2001]. Thus, for example, the verb punish has one complementation frame (because it can be complemented by a noun phrase), and the verb discover has two (because it can be complemented by a noun phrase or a sentence). Note that the verb discover has two complementation frames, but only one complement in each of these frames. Another type of information associated with the complementation frame is the number and types of thematic roles the verb assigns. Thematic roles are the roles performed by the various participants in an event, such as the agent of the event or its theme. Very generally, they describe “who did what to whom” in the sentence.

It is assumed that the complementation frames are represented in the lexical entry of the verb. The representation of this knowledge in the lexical entry is supported by various studies that have tested lexical access and online sentence processing of verbs [e.g., Boland, 1993; Boland et al., 1990; Holmes, 1987; Shapiro et al., 1987, 1993; Shetreet et al., 2007; Tanenhaus et al., 1989; Trueswell et al., 1993].

Every native speaker can easily determine the grammaticality of the above sentences. However, the way verbs with optional complements are represented in the lexicon is complicated and controversial. This is because some essential linguistic principles require the thematic structure of a verb to be fully projected across the syntactic levels of the sentence containing the verb [e.g., the Projection Principle and the theta criterion, Chomsky, 1981, 1982]. That is, each thematic role that a verb can assign must be realized by an argument: a subject or a complement. Thus, the missing complement in sentences like (3a) may appear to contradict these principles, because one thematic role is not assigned.

Different linguistic theories resolve this contradiction by assuming different lexical representations for optional verbs [Thompson and Hopper, 2001]. One approach (“the two frames theory”) argues that verbs that allow optional complements have two separate complementation frames: one with and one without a thematic role for the complement [Booij, 1992; Engelberg, 2002; Manzini, 1992; Van Valin and LaPolla, 1997]. This means that verbs with optional complements are represented like other verbs that allow two different frames (or types of complements), such as discover. However, there is a fundamental difference between discover and eat. Verbs such as discover always assign the same number of thematic roles. By contrast, according to the two frames theory, verbs with optional complements assign a different number of thematic roles for each of their complementation frames: either zero or one complement.

The alternative approach assumes that only one complementation frame is represented in the lexical entry of verbs with optional complements. Here, the verb has a single frame, which includes one complement. When such verb is inserted into a sentence without a complement, a special operation, saturation, is executed to take care of the unassigned thematic role and, consequently, allow the omission of the complement [Chierchia, 1995; Reinhart, 2000; Rizzi, 1986]. According to Rizzi (1986), saturation can be achieved by either a syntactic or a lexical operation. In the syntactic case, the missing complement is structurally realized as a phonetically null element, whereupon it receives the thematic role of the complement. In this analysis, all the sentential positions of the verb are present, like in sentences in which the complement is realized [Cummins and Roberge, 2004]. Under the lexical assumption, the missing complement is saturated by a lexical operation, an operation that takes place within the lexicon and occurs when the verb is accessed, prior to the insertion of the verb into the sentence [Bresnan, 1982; Dowty, 1978, 1989]. That is, the position of the optional complement is not structurally projected in the syntax, and the verb is inserted into the sentence without complements (see Fig. 1). When a sentence that includes a verb without a complement is heard, the parser reaches a verb without receiving an overt complement to satisfy its complementation frame, and it therefore adds the semantic content of the general complement to saturate the thematic role in the thematic frame and builds a syntactic tree of a verb without a complement (i.e., without adding a syntactic node for a null complement).

Figure 1.

Figure 1

Open in a new tab

Lexical and syntactic operations forming verbs with optional complements in sentences in which the complement is omitted.

The above approaches differ in their hypotheses regarding the number of complementation frames and the number of complements in the sentence (Table I). Obviously, the two frames theory assumes that the number of frames for verbs with optional complements is two, whereas the alternative theories assume that these verbs have only one frame. The one frame approaches differ regarding the assumed number of complements in a sentence. The syntactic saturation theory assumes that one complement exists in all sentences that include verbs with an optional complement, even when the optional complement is not realized, because it assumes a null complement in these sentences (as exemplified in Fig. 1a). The lexical saturation theory assumes no complements in this structure, because the optional complement is already saturated in the lexicon and does not appear in the syntactic structure (Fig. 1b). These distinctions are the basis for the way we propose to discriminate between the above approaches.

Table I.

The assumed number of frames and complements in the three theories for representation of optional complements, for sentences with no complement (“eat at home”) and sentences with a complement (“eat an apple”)

Theory Number of frames Number of complements
Two‐frames
  No complement 2 0
  With a complement 2 1
One‐frame
 Syntactic saturation
  No complement 1 1
  With a complement 1 1
 Lexical saturation
  No complement 1 0
  With a complement 1 1
Open in a new tab

Until now, no psycholinguistic or neurolinguistic study has directly compared these theories regarding verbs with optional complements, but much can be learned from neuropsychological studies that tested the processing of multiple frames and multiple complements, which can lay the basis for an exploration of verbs with optional complements. Studies by Shapiro and his colleagues discovered that healthy individuals and individuals with Broca's aphasia show increased reaction times for verbs with more frames [Shapiro and Levine, 1990; Shapiro et al., 1987, 1989, 1991, 1993]. However, individuals with Wernicke's aphasia do not show such pattern of sensitivity to the number of frames a verb has [Shapiro and Levine, 1990; Shapiro et al., 1993]. Also, individuals who have aphasia that originated in temporoparietal damage show impaired production and judgment of the number and types of frames [Biran and Friedmann, 2007]. Imaging studies supported the clinical evidence, showing increased cortical activation in response to increased number of frames in Wernicke's area (left superior temporal gyrus, STG), as well as areas in the left inferior frontal gyrus [IFG; Shetreet et al., 2007]. Our earlier study (Shetreet et al., in press) also showed that verbs with more frames produced greater activation in the left STG, regardless of the type of complement that they select: there was no significant difference between activations related to verbs that select a sentential complement or a noun phrase and those related to verbs that select a preposition phrase or a noun phrase, indicating that the inclusion of a sentential complement in the complementation frames of verbs did not influence posterior temporal activations associated with the number of complementation frames. These results suggest that the above areas, especially left STG, are specifically involved in the processing of multiple frames.

The picture regarding the effect of the number of complements is somewhat more complicated. The number of complements, unlike the number of frames, typically does not affect verb access: healthy individuals did not show reaction‐time differences between verbs with one complement (e.g., punish) and verbs with two complements [e.g., give as in “Marko gave Dan a present”, Shapiro et al., 1987]. Also, the cortical sites that showed sensitivity to the number of complements were the medial precuneus and the anterior cingulate [Shetreet et al., 2007], which had not traditionally been considered language areas [see Cavanna and Trimble, 2006 for a review about the precuneus and Bush et al., 2000 and Paus et al., 1998 for reviews about the anterior cingulate]. Other neuroimaging studies found involvement of other areas in the processing of the number of complements. Ben‐Shachar et al. [ 2003] found activations in left STG, but they did not report of control for the number of frames, and so it could well be that the observed results were mediated by the number of frames. In a different study, bilateral activations in the supramarginal and angular gyri were found [Thompson et al., 2007]. Their study used isolated verbs, rather than sentences, which could account for their different findings.

In this study, we use knowledge regarding the areas that participate in the processing of the number of complementation frames and the number of complements to assess the number of frames and complements in the representation of verbs with optional complements. This is done in three steps: first, we ask whether verbs with optional complements have one or two frames (one with and one without a complement). Here, we compare verbs that have a known number of frames (one or two) with verbs that have an optional complement. This allows us to adduce evidence in support of either theories that assume that the lexical representation of verbs with an optional complement includes one frame, or theories that assume that the lexical representation of these verbs includes two frames. We then examine the activations in sites that respond to the number of complements by comparing verbs with known number of complements (zero or one) to verbs with optional complements (without a complement). These comparisons could enable a differentiation between theories that assume that sentences in which the optional complement is omitted do not contain complements (lexical saturation) and theories that assume that these sentences include a (null) complement (syntactic saturation). Finally, we try to localize regions that have a role in the processing of optional complements.

METHODS

Participants

Nineteen healthy volunteers (eight females) aged 22–44 (mean age, 29.2) participated in the experiment. They had normal hearing, no language impairment, and no psychiatric or neurological history. All participants were native speakers of Hebrew, which was their sole mother tongue. They were all right handed [as assessed by the Edinburgh Handedness Inventory [Oldfield, 1971]]. Written informed consent was obtained from all participants. The Tel‐Aviv Sourasky Medical Center and Tel Aviv University ethics committees approved the experimental protocol.

Materials and Procedure

To select the verbs, five linguists and psycholinguists judged Hebrew verbs according to the optionality of their complements. On the basis of their judgments, we chose 32 verbs, eight verbs for each of four categories. One category included verbs with optional complement, each of which was used in two forms: the full form in which the complement was realized, and the omitted form in which the complement was not realized. The other categories were verbs with one frame of no complements (intransitives, like sneeze), verbs with one frame of one obligatory complement (obligatory transitives, like punish), and verbs with two frames of one obligatory complement (like discover). Thus, five conditions were included in this study.

Each verb appeared in four different sentences, except for the verbs with optional complements, which appeared in eight different sentences, four in the full form, and four in the omitted form, giving a total of 160 sentences for the entire experiment (five conditions × eight verbs per condition × four sentences per verb). The sentences included four constituents: a subject (a person's name), the verb, a complement or an adjunct (according to the verb type), and an adjunct or a modifier1 (Table II). Complements were either noun phrases or prepositional phrases. In all the sentences, the verb was inflected for third person singular and past tense. In each condition, half of the sentences included a feminine subject (and hence the verb was inflected for the feminine), and half included a masculine subject. The structure of the complements and adjuncts was controlled across conditions. The verbs were controlled for frequency as determined by the Hebrew Word Frequency Database [Frost and Plaut, 2005]. ANOVA showed no significant differences among the various conditions (F 3,28 = 0.79, P = 0.5 for the combined frequency of masculine and feminine forms; mean frequency = 15.5, 22.7, 8.3, and 24.8, respectively, for verbs with no complements, verbs with one obligatory complement, verbs with an optional complement, and two‐frame verbs). The number of syllables in each sentence was controlled and ranged between 10 and 13 syllables (mean = 12.5 syllables in each block).

Table II.

Example sentences of each condition

Condition Examples
No complements Dana pihaka [etmol] [b‐a‐miklaxat]
Dana yawned [yesterday]adjunct [in‐the‐shower]adjunct
Optional complement (omitted form) Ziv axal [etmol] [b‐a‐mis'ada]
Ziv ate [yesterday]adjunct [in‐the‐restaurant]adjunct
Optional complement (full form) Sara axla [et ha‐tapuax] [b‐a‐xacer]
Sara ate [the‐apple]complement [in‐the‐yard]adjunct
Obligatory complement Ron shavar [et ha‐kos] [b‐a‐xatuna]
Ron broke [acc the‐glass]complement [in‐the‐wedding]adjunct
Two frames Gal siyma [et ha‐sefer] [b‐a‐xacot]
Gal finished [acc the‐book]complement [in‐the‐midnight]adjunct
Open in a new tab

Sentences were divided into 40 blocks that were presented in two separate runs. Each block consisted of four sentences with verbs of the same condition. Each condition was repeated eight times, four blocks in each run. Each verb appeared in a block only once. In each block, half of the sentences included a subject and a verb in a masculine form and half in a feminine form. The blocks and the sentences in each block were presented in a pseudo‐random order (which was determined for each participant using a Matlab script), with no more than two consecutive blocks of the same condition. The order of the blocks within a run as well as the order of the runs was counter‐balanced across participants, thus in some runs the omitted form was presented first, whereas in other runs the full form was presented first. The presentation of each block lasted 14 s. Sentences were separated by silence periods of 1500 ms. A tone was heard at the end of each block to signal 6 or 8 s of silence. During silence, participants were instructed to concentrate on the noises of the MRI scanner. Stimuli were delivered to the participants via MRI compatible headphones using Presentation software (http://nbs.neuro-bs.com).

Throughout the experiment, participants had to perform a comprehension task to ensure that they attended to the sentences and processed them fully. In this task, the participants were asked to listen to the sentence and decide whether the event described in the sentence was more likely to happen at home or not. Responses were given during the intrablock silences (responses were not allowed before the end of a sentence or after the beginning of the following sentence). Participants were requested to press a “yes” or a “no” button with their left hand fingers (to avoid interference in frontal language areas) after the end of the sentence. For example, for the sentence “Dan slept in the yellow tent,” participants had to press the “no” button; for the sentence “Jane yawned in bed at midnight,” they had to press the “yes” button. There were equal numbers of predicted “yes” and “no” responses in the entire experiment, and the number of predicted “no” responses in each block differed (ranging one to three). All responses and reaction times were recorded.

Each subject completed a short practice outside and inside the MRI scanner. The four practice blocks included sentences that were similar to those used in the experiment, but with verbs that were not included in the experiment. Each session lasted 8 min and the entire imaging session (including practices, anatomical, and other functional scans) lasted approximately an hour and a half.

Data Acquisition

MRI scans were conducted in a whole‐body 3 Tesla, General Electric scanner, located at the Whol Institute for Advanced Imaging in the Tel‐Aviv Sourasky Medical Center. Anatomical images for each subject were acquired using a 3D spoiled gradient echo (SPGR) sequence with high resolution to allow volume statistical analyses in single participants. The whole brain was covered by 150–166 slices, 1‐mm thick (no gap). Functional MRI protocols included T2*‐weighted images in runs of 363 volumes acquired in two separate functional sessions. We selected 33 sagital slices (based on a mid‐sagital slice), 3.5‐mm thick (no gap), covering the whole of the cerebrum and most of the cerebellum. We used FOV of 20 cm and matrix size of 64 × 64, TR = 2,000 ms, TE = 30, and flip angle = 90.

Data Analysis

Image analysis was performed using SPM2 (Wellcome Department of Cognitive Neurology, http://www.fil.ion.ucl.ac.uk/spm/). Functional images from each subject were motion‐corrected, normalized to the SPM EPI template, resampled with a voxel size of 3 × 3 × 3 mm [Ashburner and Friston, 1999], and spatially smoothed using a Gaussian filter (6‐mm kernel). The analysis assumed the general linear model [GLM, Friston et al., 1995] as implemented in SPM2. The BOLD response was modeled with the canonical hemodynamic response function. Head motion parameters were added as regressors [Friston et al., 1995]. An additional regressor modeled the button presses using the end of each sentence for the onsets (because responses were not allowed before this time) and 1,500 ms for the duration (because this was the time allocated for responses). In the group level analysis, contrasts were examined using t‐test with uncorrected P value (P < 0.001) and a minimum cluster size of 10 voxels, unless stated otherwise. The peak coordinates were transformed to Talairach space using mni2talsoftware (http://www.mrc-cbu.cam.ac.uk/Imaging/mnispace.html) and areas of activations, as well as Brodmann areas (BA), were localized using Talairach daemon software [Lancaster et al., 1997].

RESULTS

Two Frames Versus One Frame

First, we focus on the number of complementation frames that optional verbs select, which describes how many different types of complementation options a verb has. To determine whether verbs with optional complements have one or two frames, we undertook two types of analyses. In the first analysis, we compared verbs with optional complements to verbs with a definite number of complementation frames, that is, one frame (like punish or sneeze) or two frames (like discover). In our reasoning, if the differences between optional verbs and two‐frame verbs were greater than the differences between optional verbs and one‐frame verbs ([discover > eat] > [eat > punish]), this would imply that verbs with optional complements are likely to have one frame (given that the different types of verbs are controlled for many attributes other than the number of frames). The opposite pattern ([eat > punish] > [discover > eat]) would imply that optional verbs are likely to have two frames (see Fig. 2).

Figure 2.

Figure 2

Open in a new tab

Illustration of the possible results regarding the number of frames of optional verbs and the reasoning that was to be applied to draw conclusions from these results.

In addition, we also considered the cortical regions found in the comparison between two‐frame and one‐frame verbs in relation to the cortical regions found in the comparisons of optional verbs and other verbs. The rationale was that the cortical locations identified in the comparison of two‐frame verbs and one‐frame verbs were the ones related to the processing of more complementation frames. Hence, if the locations found in the comparison between two‐frame verbs and optional verbs were also identified as the locations related to the number of frames ([discover > punish] ∼ [discover > eat]), this would imply that optional verbs differ from two‐frame verbs in the same way that one‐frame verbs differ from two‐frame verbs, that is, in the number of frames. This would indicate that verbs with optional complements have only one frame of complementation. If, on the other hand, the locations found in the comparison between optional verbs and one‐frame verbs would appear in the number‐of‐frames regions ([discover > punish] ∼ [eat > punish]), this would imply that verbs with optional complements resemble verbs with two frames and, therefore, would indicate that the lexical entry of verbs with optional complements includes two separate complementation frames. Note that we did not expect the results of the above comparisons to be completely identical, because, naturally, optional verbs differed from the other verbs in their complement optionality. Nonetheless, many properties of the different verb types, other than the number of frames, were controlled and matched, and therefore a broad similarity was expected. Thus, we examined the cortical activations of the different verbs in an attempt to identify activations that show greater similarity to the cortical activations associated with verbs that have optional complements.

Results: Two frames vs. one frame

We first compared optional verbs to one‐frame verbs, which yielded no differential activations: optional verbs in the full form were compared to verbs with one obligatory complement, and optional verbs in the omitted form were compared with no‐complement verbs (full optional > one frame‐one complement; one frame‐one complement > full optional; omitted optional > one frame‐no complements; and one frame‐no complements > omitted optional). By contrast, several activations were revealed when comparing optional verbs (both full and omitted forms) with two‐frame verbs (Table III and Fig. 3). The comparison of full optional verbs and two‐frame verbs (two frames > full optional) revealed activations in the left STG (BA 40/22/39), left middle temporal gyrus (MTG; BA 21), and left middle frontal gyrus (MFG; BA 6). The other comparison, of omitted optional verbs and two‐frame verbs (two frames > omitted optional), showed activations in left STG (BA 40/22/39), left MFG (BA 6), left inferior temporal gyrus (ITG; BA 21), left superior frontal gyrus (SFG; BA 9), left MTG (BA 22), right MFG (BA 6), left IFG (BA 44/47), right IFG (BA 44/47 and BA 45), medial precuneus, and right MTG/ITG. The difference between the absence of activations in the comparisons of optional verbs to one‐frame verbs and the presence of many activations in the comparisons of optional verbs with two‐frame verbs suggests that optional verbs are more similar to one‐frame verbs than to two‐frame verbs, and by inference have one complementation frame.

Table III.

Coordinates, cluster sizes, and maximal t values of brain regions identified in the comparisons of the number of complementation frames (P < 0.001, cluster size > 10)

Region x y z Number of voxels in cluster t max
Two frames > full form
 Left STG (BA 40/22/39) −53 −46 16 14 4.19
 Left MTG (BA 21) −53 −9 −15 39 5.66
 Left MFG (BA 6) −36 2 44 13 5.48
Two frames > one frame
 Left STG (BA 40/22/39) −59 −59 24 63 5.02
Two frames > omitted form
 Left temporal 494
  Left STG (BA 40/22/39) −53 −49 12 5.36
  Left ITG (BA 21) −59 −7 −15 8.61
 Left MFG (BA 6) −36 5 47 143 5.24
 Left SFG (BA 9) −15 8 34 137 6.2
 Right MTG (BA 21) 56 −46 8 32 6.26
 Right MFG (BA 6) 36 5 44 14 4.11
 Left IFG (BA 44/47) −48 26 −4 265 7.92
 Right IFG (BA 45) 59 21 18 65 4.75
 Right IFG (BA 44/47) 33 23 −6 52 4.49
 Medial precuneus −9 −45 35 32 4.56
 Right MTG/ITG 50 −10 −20 78 6.08
Two frames > one frame
 Left STG (BA 40/22/39) −42 −48 27 165 5.83
 Left MFG (BA 6) −33 2 47 86 5.99
 Left ITG (BA 20) −50 −10 −22 45 5.93
 Left IFG (BA 9) −50 19 21 154 5.28
 Left SFG (BA 9) −12 54 36 26 4.4
 Left MTG (BA 22) 56 −44 2 19 4.43
 Anterior cingulate (BA 32) −6 22 40 184 7.01
 Right STG (BA 22) 56 −43 13 11 4.47
Open in a new tab
Figure 3.

Figure 3

Open in a new tab

Regions activated in the comparisons aimed to identify the number of complementation frames that are represented in the lexical entry of verbs with optional complements (P < 0.001, cluster size > 10). The left STG is activated in all of these comparisons. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

In addition, we also compared the locations identified in the comparison of optional verbs and two‐frame verbs to the locations identified in the comparison between one‐frame and two‐frame verbs (Table III and Fig. 3). The contrast between two‐frame verbs and one‐frame verbs (two frames > one frame‐one complement) revealed a single cluster of activations in the left STG (BA 40/22/39). A similar activation, in the left STG, was identified in the comparison of two‐frame and optional verbs (two frames > full optional), implying similarity between verbs with one frame and verbs with optional complements. The contrast between two‐frame verbs and one‐frame verbs that have no complement (two frames > one frame‐no complement) revealed activations that were identified in the comparison of two‐frame verbs and omitted optional verbs: left STG (BA 40/22/39), left MFG (BA 6), left ITG (BA 20), left SFG (BA 9), and left MTG (BA 22). Additional areas that were identified in the comparison between two‐ and one‐frame verbs included the anterior cingulate (BA 32), the left IFG (BA 9), and right STG (BA 22). Thus, in these comparisons too, many cortical activations were identified both in the comparison of two‐frame and one‐frame verbs and in the comparison of two‐frame and optional verbs, suggesting that verbs with optional complements differ from two‐frame verbs in a similar way to the way that one‐frame verbs differ from two‐frame verbs. Hence, our results support the idea that optional verbs have one complementation frame.

Interim Discussion: How Many Frames?

Our most robust results were that, whereas the comparisons of optional verbs to one‐frame verbs did not show any activation, the comparisons to two‐frame verbs showed clear activations spanning diverse cortical areas. This difference, given that the verbs were controlled for many properties other than the number of frames (and complements), led us to assume that verbs with optional complements and two‐frame verbs had different number of complementation frames, that is, that verbs with optional complements had one complementation frame. Additionally, the results of the comparisons between two‐frame verbs and verbs with optional complements (full form and omitted form) showed activations in similar location to the locations of the comparison between two‐frame verbs and one‐frame verbs (one complement or no complements, respectively). This similarity further suggested that verbs with optional complements resembled one‐frame verbs. Moreover, the left STG was the only area that was activated in all of the comparisons between two‐frame verbs and one‐frame verbs (including verbs with optional complements). This area has repeatedly been associated with the processing of multiple complementation frames [Shapiro and Levine, 1990; Shapiro et al., 1993; Shetreet et al., 2007]. In particular, this region is activated when verbs with many frames are compared with verbs with fewer frames. We see these results as supportive of the idea that the representation of verbs with optional complements includes one complementation frame, and as an argument against the idea that such representation includes two frames.

Syntactic Operation Versus Lexical Operation

After setting up the number of complementation frames that verbs with optional complements select, we focus on their number of complements. This attribute allows us to discriminate between two theories, both of which assume a single frame but nevertheless differ with respect to the nature of the operation that allows the complement omission: syntactic saturation and lexical saturation. According to the syntactic approach, the syntactic structure of sentences in the omitted form resembles that of sentences in the full form, as well as sentences containing verbs with an obligatory complement, because a null complement is assumed to exist in the sentence. By contrast, the lexical approach implies that sentences in the omitted form resemble sentences containing verbs with no complements, because both are inserted from the lexicon into the sentence already without a complement.

Therefore, we compared sentences with optional verbs in the omitted form to sentences with verbs whose number of complements was known, that is, verbs with no complements (e.g., sneeze) and verbs with one complement (e.g., punish), including optional verbs in their full form. The rationale was that if the differences between one‐complement verbs and optional verbs in the omitted form were more robust (i.e., showed more activations) than the differences between no‐complement verbs and omitted optional verbs ([punish > eat [omitted]] > [eat [omitted] > sneeze]), this would imply that optional verbs are likely to have no complements. The opposite pattern ([punish > eat [omitted]] < [eat [omitted] > sneeze]) would imply that optional verbs are more similar to verbs with one complement, and thus, are likely to have one complement (see Fig. 4).

Figure 4.

Figure 4

Open in a new tab

Illustration of the possible results regarding the number of complements of optional verbs in the omitted form and the reasoning that was to be applied to draw conclusions from these results.

We also compared the locations identified in the comparison between verbs with one obligatory complement and verbs with no complements, and the locations revealed in the comparisons of optional verbs to one‐ and no‐complement verbs. Here, if the locations identified in the comparison between one‐ and no‐complement verbs were similar to the locations identified in the comparison between one‐complement verbs and omitted optional verbs ([punish > sneeze] ∼ [punish > eat [omitted]]), this would imply that the omitted optional verbs differ from one‐complement verbs and resemble no‐complement verbs, indicating that there are no complements in the omitted form and that a lexical operation is executed. The opposite pattern, in which the locations identified when comparing one‐complement and no‐complement verbs were similar to the locations identified when comparing the omitted optional verbs and no‐complements verbs ([punish > sneeze] ∼ [eat [omitted] > sneeze]), would imply that the omitted form resembles one‐complement verbs and therefore would indicate that the omitted form includes one (null) complement and that a syntactic operation takes place. Here too, we did not expect complete identity of the activations associated with the different verb types, which differ mainly in complement optionality. Nonetheless, we did expect some similarity between optional verbs and another group of verbs with regard to the number of complements, as other properties of the verbs were controlled across the different types of verbs.

Results: Syntactic operations vs. lexical operation

Comparing omitted optional verbs with no‐complement verbs (omitted optional > no complements) revealed no activations. By contrast, the comparison of one‐complement verbs with omitted optional verbs (one complement > omitted optional) showed several activations in the medial precuneus, the left MFG (BA 6), posterior cingulate, right inferior parietal lobule (BA 40), left and right anterior MTG (BA 21), right MFG (BA6), left posterior MTG, and left and right IFG (BA 45 and BA 47). An important comparison between the full form and the omitted form (full optional > omitted optional), which included the same verbs in different sentential contexts, also revealed activations in the medial precuneus, left MFG (BA 6), left superior parietal lobule, right MFG (BA 8), and left IFG (BA 45 and BA 47). The difference between the extent of activations in the comparisons of omitted optional verbs and one‐complement verbs and the extent of activations in the comparison of omitted optional verbs and no‐complement verbs suggests that omitted optional verbs differ from verbs with one complement and thus indicates that they have no complements.

In addition, we also compared verbs with one obligatory complement to no‐complement verbs (one complement > no complement) and identified activations in the medial precuneus, and left MFG (BA 6), as well as posterior and anterior cingulate (BA 32). Thus, similar activations, mainly in the medial precuneus and left MFG, were identified in the comparison of one‐complement verbs to both omitted optional verbs and no‐complement verbs, suggesting that omitted optional verbs resemble no‐complement verbs. The results of the above comparisons are shown in Table IV and Figure 5.

Table IV.

Coordinates, cluster sizes, and maximal t values of brain regions identified in the comparisons of the number of complements (P < 0.001, cluster size > 10)

Region x y z Number of voxels in cluster t max
One complement > omitted form
 Medial precuneus 6 −51 36 20 3.61
 Left MFG (BA 6) −42 8 47 47 5.97
 Right inferior parietal lobule (BA 40) 39 −50 41 35 7
 Left MTG (BA 21) −59 −4 −12 26 6.45
 Right MTG (BA 21) 59 −4 −12 36 5.37
 Right MFG (BA6) 36 14 52 14 4.91
 Left posterior MTG −62 −35 −3 66 4.93
 Medial posterior cingulate (BA 23) 0 −55 17 25 4.9
 Left IFG (BA 45) −53 21 18 35 5.07
 Left IFG (BA 47) −50 20 −6 36 4.67
 Right IFG (BA 45) 53 18 13 20 4.75
 Right IFG (BA 47) 45 29 −12 11 3.76
Full form > omitted form
 Medial precuneus 0 −57 19 173 6.91
 Left MFG (BA 6) −39 8 47 80 6.37
 Left‐superior parietal lobule −33 −71 42 53 6.35
 Right MFG (BA 8) 3 46 39 16 5.12
 Left MTG (BA 21) −65 −38 −8 19 4.75
 Left IFG (BA 45) −50 18 16 16 4.52
 Left IFG (BA 47) −45 20 −6 13 4.17
One complement > no complement
 Medial precuneus −3 −65 36 42 4.29
 Anterior cingulate (BA 32) 3 25 40 121 6.96
 Posterior cingulated −3 −54 19 40 6.64
 Left MFG (BA 6) −39 8 47 23 4.95
Open in a new tab
Figure 5.

Figure 5

Open in a new tab

Regions activated in the comparisons aimed to identify the number of complements in the omitted form of verbs with optional complements (P < 0.001, cluster size > 10). The medial precuneus was identified in all of these comparisons. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Interim Discussion: How Many Complements?

The comparisons between optional verbs without an overt complement and verbs with one complement (optional or obligatory) yielded clear activations, whereas no differential activation was found in the comparison between optional verbs without an overt complement and verbs with no complements. Furthermore, the comparison between the same optional verbs when they appeared in different sentential contexts, with or without a realized complement, revealed extensive activations, even though the verbs in the sentences were identical. Had there been a (null) complement in the sentence, we would expect optional verbs in the omitted form to behave like one‐complement verbs, but instead they showed activations that were similar to the activations related to verbs with no complements. Because verb properties other than the number of complements were carefully controlled, these findings suggest that verbs with omitted optional complements have a similar number of complements as verbs with no complements, i.e., zero.

This is also supported by the similarity detected between the locations identified in the comparison between optional verbs in the omitted form and verbs with one obligatory complement and the locations relating to the number of complements (i.e., the ones identified in the comparison between verbs with one obligatory complement and verbs with no complements). Both in the current study and in our previous study [Shetreet et al., 2007] of the effect of the number of complements on brain activations, the medial precuneus was active when sentences with one complement were compared to sentences without complements, including sentences with omitted optional verbs. This area has previously been linked to lexical processing of information of a relatively high complexity [Shetreet et al., 2009a] and, specifically, to the processing of the number of complements [Den Ouden et al., 2009; Shetreet et al., 2007]. Inasmuch as activation in the precuneus reflects the number of complements, its activation in the comparison between sentences with one complement and sentences with omitted optional verbs, together with the absence of activation in this area in the comparison between verbs with no complements and optional verbs in the omitted form, suggest that a syntactic operation that inserts a null complement to the complement position is unlikely to be performed. The findings are rather consistent with the idea of lexical saturation, which does not imply the addition of a (null) complement.

Is it possible that activation related to the omitted complement has not been found because this element is null? We see this as unlikely, given that null elements do affect brain activations in psycholinguistic studies [Clahsen and Featherson, 1999; Friedmann et al., 2008; Nicol and Swinney, 1989; Tanenhaus et al., 1989; Zurif et al., 1993] and in ERP studies [Featherson et al., 2000; Fiebach et al., 2001; Kluender and Kutas, 1993]. Furthermore, in a recent neuroimaging study [Shetreet et al., 2009b], we observed brain activations related to a phonetically null element in the complement position. This was found in passive sentences, which include a trace in the complement position, after the movement of the complement to the subject position [Chomsky, 1981; Grodzinsky, 1995; Lasnik, 2002]. The comparison of sentences that include verbs with one complement to passive sentences showed no activation in the precuneus, whereas comparing sentences with one‐complement verbs to sentences with no‐complement verbs revealed activation in that area [Shetreet et al., 2009b]. This suggests that a null element that is present in the complement position leads to activations in the precuneus, which resembles the activation caused by a phonetically realized element [Shetreet et al., 2007]. These findings reduce the probability that the observed similarity of activation between optional verbs and no‐complement verbs is an artifact of the current paradigm.

Localizing Lexical Saturation

The above analysis of the data supports the idea that the omitted form of verbs with optional complements is realized using lexical saturation, and so now we try to identify the regions that are involved in performing this operation. The operation takes place when verbs that allow optional complements are inserted into sentences without a complement (such as “John ate”). Therefore, it is present only in sentences in the omitted form. We compare these sentences to their syntactic matches—sentences that have a similar syntactic structure, but a different number of complements in the lexical entry of their embedded verbs, namely sentences containing verbs with no complements (e.g., sneeze). We also compare sentences in the omitted form to their lexical matches—sentences that include verbs with the same number of complements in their lexical representation prior to the operation (i.e., one), but different syntactic structures, namely sentences in the full form and sentences containing verbs with one obligatory complement (e.g., punish). In addition, we perform a conjunction analysis of the contrasts that compare the omitted form to any of the other conditions. This is done in order to identify areas that are active in all of the comparisons. Finally, we compare the omitted form to a combination of its syntactic and lexical matches.

Results: Localizing lexical saturation

First, we compared sentences in the omitted form (which involved a lexical operation) to sentences that had a similar syntactic structure, namely, sentences with no complements (omitted form > no complements). This comparison yielded no activation using a threshold of P < 0.001, cluster size > 10, but using a more relaxed threshold of P < 0.005, cluster size > 20, we found an activation in the left temporo‐parieto‐occipital (TPO) junction (BA 39/19) (Table V). The left TPO junction (BA 39/19), as well as the left fusiform gyrus (FG; BA 20/37), were activated when comparing sentences in the omitted form with both their lexical matches (which had a similar lexical representation): sentences in the full form (omitted form > full form), and sentences with an obligatory complement (omitted form > obligatory complement) (Table V). One area was activated only in the latter comparison, the left inferior parietal lobule (BA 40). A triple conjunction of each of these comparisons was conducted. At both strict and more liberal thresholds (P < 0.001, cluster size > 10; P < 0.005, cluster size > 10; P < 0. 01, cluster size > 10), no activations were found. Because conjunction is considered a rather conservative analysis [Nichols et al., 2005], we also used the very liberal threshold of P < 0.05 (cluster size > 40), and found activations only in three areas: the left TPO junction (BA 39/19), the left FG (BA 20/37), and the left post central gyrus (BA 2) (Table V and Fig. 6).

Table V.

Coordinates, cluster sizes, and maximal t values of brain regions identified when comparing sentences in which the lexical operation takes place with sentences in which it does not (P < .001, cluster size > 10, unless stated otherwise)

Region x y z Number of voxels in cluster t max
Omitted form > no complements *
 Left TPO junction (BA 39/19) −45 −64 24 22 3.41
Omitted form > full form
 Left TPO junction (BA 39/19) −34 −58 9 11 4.95
 Left FG (BA 20/37) −33 −42 −21 20 5.79
Omitted form > one complement
 Left TPO junction (BA 39/19) −45 −81 18 18 5.56
 Left FG (BA 20/37) −33 −42 −16 28 5.35
 Left inferior parietal lobule (BA 40) −62 −30 37 32 5.69
Omitted form > no complements; omitted form > full form; omitted form > one complement **
 Left TPO junction (BA 39/19) −27 −47 −13 48 3.38
 Left FG (BA 20/37) −36 −61 6 48 2.75
 Left postcentral gyrus (BA 2) −62 −22 29 78 2.54
Omitted form > no complements + one complement
 Left FG (BA 20/37)
Omitted form > no complements + full form *
 Left FG (BA 20/37) −35 −42 −19 28 4.27
 Left post central gyrus (BA 2) −62 −22 29 20 2.88
Omitted form > no complements + full form + one complement
 Left FG (BA 20/37) −33 −42 −16 21 5.53
Open in a new tab
*

P < 0.005, cluster size > 20.

**

P < 0.05, cluster size > 40.

Figure 6.

Figure 6

Open in a new tab

Regions activated when the lexical operation takes place. The conjunction analysis (P < 0.05, cluster size > 40) showing activations in the left FG (top), the left TPO junction (bottom), and the left post central gyrus. The combined analysis (P < 0.001, cluster size > 10) showing activation in the left FG. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

We then compared sentences in the omitted form to sentences without an operation: a combination of their syntactic and lexical matches to identify the regions that were involved in the lexical operation (Table V). Comparing omitted optional verbs to verbs with no complements and verbs with an obligatory complement (omitted form > no complements + obligatory complement) showed activations in the left FG (BA 20/37), whereas comparing omitted optional verbs to verbs with no complements and the full form (omitted form > no complements + full form) revealed no activations. Using a more relaxed threshold (P < 0.005, cluster size > 20), this comparison showed activations in the left FG (BA 20/37), as well as the left post central gyrus (BA 2). When comparing the omitted form to all other three conditions (omitted form > no complements + full form + obligatory complement), only the left FG (BA 20/37) was activated (see Fig. 6).

Interim Discussion: Where Is the Operation?

A few areas have been identified when comparing sentences that include optional verbs without an overt complement, where a lexical saturation operation is assumed, to sentences that do not require lexical saturation. One such area appeared in almost all of the above comparisons, namely, the left fusiform gyrus. This area was found in all of the comparisons that used the strict threshold of P < 0.001 (omitted optional > full optional; omitted optional > one complement; omitted optional > one complement + no complement; and omitted optional > one complement + full optional + no complement), as well as the combined comparison of optional verbs in the omitted form to verbs with no complements and the full form and which was conducted at a more relaxed threshold of 0.005. It was absent only from the comparison between verbs with optional complements in the omitted form and verbs with no complements.2 Thus, bearing in mind the relaxed thresholds that were used in some of the comparisons, the results seem to support the indication that FG is involved in executing the lexical operation.

Another area identified in two of the comparisons of individual conditions (as well as in the third one with a less strict threshold of 0.005) was the TPO junction. This area was also identified in the conjunction analysis, but not in the combined comparisons. This leaves an open possibility that the left TPO also takes part in the processing of lexical saturation.

GENERAL DISCUSSION

This study used fMRI to constrain the linguistic arguments regarding the representation of optional complements. Among the three approaches that were examined here—the two‐frame approach, the syntactic saturation approach, and the lexical saturation approach—our results seem to provide support for lexical saturation. We base this inference on the number of complementation frames and the number of complements that are required by verbs with optional complements. Whereas no activations were found when these verbs were compared to one‐frame verbs, diverse activations were revealed in the comparison with two‐frame verbs. These results suggest that verbs with optional complements are more different from two‐frame verbs than from one‐frame verbs. Moreover, many of the locations identified when comparing optional verbs to two‐frame verbs revealed activations that were also identified when comparing one‐frame verbs and two‐frame verbs, namely, areas that are sensitive to the number of frame differences. Among the areas that were identified in all of the comparisons was the left STG, which was found to be involved in the processing of multiple frames [Shapiro and Levine, 1990; Shapiro et al., 1993; Shetreet et al., 2007]. This suggests that verbs with optional complements have only one frame.

With respect to the representation of optional verbs that appear in a sentence without a complement, our results support the idea that verbs with optional complements in the omitted form have no (null) complements in their syntactic structure. This is derived from the absence of activations in their comparison to verbs with no complements and the diverse activations in their comparison to verbs with one complement. Such pattern suggests that verbs with optional complements in the omitted form are probably similar to verbs with no complements. Additionally, many of the activated areas found in the comparison of verbs with omitted optional complements and verbs with one complement were also found in the comparison between verbs with no complement and verbs with one complement. These activations included the medial precuneus, which was previously linked to the processing of multiple complements [Den Ouden et al., 2009; Shetreet et al., 2007], indicating that omitted optional verbs and one‐complement verbs differ in the relevant respect—the number of complements.

Hence, the results suggest that the lexical entry of verbs with optional complements includes only the representation of the full form. The omitted form of these verbs seems to be generated by an operation that saturates the thematic role of the complement in the lexicon and, therefore, the complement is not projected into the syntactic structure.

The lexical nature of the formation of optional complements was also displayed in the comprehension of sentences that require reanalysis (garden path sentences) by individuals with conduction aphasia [Friedmann and Gvion, 2007]. The performance of these patients in understanding garden path sentences with optional complements resembled their performance in understanding lexical garden path sentences, only required re‐access to the lexical representation of a specific word (to reactivate its multiple meanings) and differed from the understanding of syntactic garden path sentences, which only required syntactic reanalysis. This suggests that the reanalysis of optional verbs—from a representation with a complement to a representation without a complement—involved reaccessing the lexicon.

The current study also used fMRI in the “traditional” manner, namely, in an attempt to locate the cortical regions that were involved in the operation of lexical saturation. In our results, the best candidate for this process was the left FG, which was identified in almost all comparisons between sentences in which the operation took place (i.e., sentences in the omitted form) and sentences in which no such operation was involved. This is in line with previous finding according to which the FG plays a role in lexical processing due to its activation during lexical decision [Poeppel et al., 2004]. In addition, it is frequently associated with semantic or lexical‐semantic processing. Repeated activations in the FG were observed in imaging studies during semantic decision tasks in diverse modalities, e.g., visual [Büchel et al., 1998; Wagner et al., 1998], tactile [for blind individuals, Büchel et al., 1998], and auditory [Balsamo et al., 2006; Binder et al., 1997; Chee et al., 1999]. Furthermore, ERP recordings from the left temporal lobe showed that the FG was sensitive to words (rather than nonwords) as well as to their semantic context [Nobre et al., 1994].

Another region that may subserve lexical saturation is the left TPO junction (BA 39/19), which was identified in all the comparisons of the omitted form with each of the other conditions individually (one of which used a slightly relaxed threshold), but not in the combined comparisons. This area was found to be activated in lexical decision [Cappa et al., 1998] as well as in semantic decision tasks and with different modalities [Mummery et al., 1998; Vandenberghe et al., 1996]. It was also activated during verb generation [Kemeny et al., 2006; Warburton et al., 1996], an activation that was attributed to semantic processing by Warburton et al. [ 1996].

The semantic involvement of the FG and the TPO junction implies that the lexical operation that generates the omitted form of verbs with optional complements may have a semantic source. Some researchers have argued that the representation of verbs with optional complements includes certain lexical marking on the complements that can be omitted [Fillmore, 1986], whereas others suggest a semantic explanation [Groefsema, 1995; Rizzi, 1986]. According to the semantic account, the conceptual representation of a verb specifies the semantic types of complements that the verb can appear with, such as thing, place, or event. Some verbs put more specific semantic restrictions on their complements. For example, the verb eat specifies that the type of thing is food. That is, verbs that allow complement omission select their complements from a narrow range of possible semantic categories and thus allow the omitted complements to be recovered from the verb [Fellbaum, 1999; García Velasco and Portero Muñoz, 2002; Resnik, 1996; Rice, 1998]. Resnik [ 1996] examined the semantic constraints of several verbs and found that some verbs (e.g., drink, eat, or pour) had strong selectional preference. By contrast to verbs with optional complements, verbs such as accept, object, or wait allow complement omission only when the complement could be recovered from the context. Thus, it is possible that complement omission for verbs with optional complements is indeed semantic in nature.

To conclude, our findings support the idea that verbs with optional complements are represented in the lexicon with one complementation option. This study also provides suggestive evidence that the omission of the optional complement is enabled by a lexical operation. Finally, we identify the left FG (and possibly the left TPO junction) as a likely candidate to be involved in the execution of this operation. We believe that the importance of this study is not merely in this novel finding, but also in the nontraditional usage of fMRI. By contrast to the usual use of fMRI for the cortical localization of cognitive processes, this study used the putative functions of the areas that we identified in order to constrain cognitive theory and derive evidence in support of a particular linguistic approach to the representation of verbs.

Footnotes

1

Adjunct stands for any phrase that is not required by the verb and can be attached to every verb, usually in the form of time and place descriptions, for example, “at the wedding” in the sentence “John broke the glass at the wedding.” Modifier stands for any phrase added to a complement, such as “yellow” in the sentence “John touched the yellow shirt.”

2

The left FG was also seen in the conjunction of this comparison with the other individual comparisons (which was conducted at the very liberal threshold of 0.05 uncorrected) that aimed to identify areas that were activated in all the comparisons of individual conditions. This suggested that the left FG was also activated in the comparison between omitted optional verbs and no‐complement verbs, but with less magnitude.

REFERENCES

  1. Ashburner J,Friston KJ ( 1999): Nonlinear spatial normalization using basis functions. Hum Brain Mapp 7: 254–266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Balsamo LM,Xu B,Gaillard WD ( 2006): Language lateralization and the role of the fusiform gyrus in semantic processing in young children. NeuroImage 31: 1306–1314. [DOI] [PubMed] [Google Scholar]
  3. Ben‐Shachar M,Hendler T,Kahn I,Ben‐Bashat D,Grodzinsky Y ( 2003): The neural reality of syntactic transformations—Evidence from fMRI. Psychol Sci 14: 433–440. [DOI] [PubMed] [Google Scholar]
  4. Binder JR,Frost JA,Hammeke TA,Cox RW,Rao SM,Prieto T ( 1997): Human brain language areas identified by functional MRI. J Neurosci 17: 353–362. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Biran M,Friedmann N ( 2007): Lexical‐syntactic information: Intact representation and possible impairments. Lang Brain 6: 18–29. [Google Scholar]
  6. Boland J ( 1993): The role of verb argument structure in sentence processing: Distinguishing between syntactic and semantic effects. J Psycholinguist Res 22: 133–152. [Google Scholar]
  7. Boland JE,Tanenhaus MK,Garnsey SM ( 1990): Evidence for the immediate use of verb control information in sentence processing. J Mem Lang 29: 413–432. [Google Scholar]
  8. Booij G ( 1992): Morphology, semantics, and argument structure In: Roca I, editor. Thematic Structure, Its Role in Grammar. Dordrecht: Foris; pp 27–50. [Google Scholar]
  9. Bresnan J ( 1982): Polyadicity In: Bresnan J, editor. The Mental Representation of Grammatical Relations. Cambridge, MA: MIT Press; pp 149–172. [Google Scholar]
  10. Büchel C,Price C,Friston K ( 1998): A multimodal language region in the ventral visual pathway. Nature 394: 274–277. [DOI] [PubMed] [Google Scholar]
  11. Bush G,Luu P,Posner MI ( 2000): Cognitive and emotional influences in anterior cingulate cortex. Trends Cogn Sci 4: 215–222. [DOI] [PubMed] [Google Scholar]
  12. Cappa SF,Perani D,Schnur T,Tettamanti M,Fazio F ( 1998): The effects of semantic category and knowledge type on lexical‐semantic access: A PET study. NeuroImage 8: 350–359. [DOI] [PubMed] [Google Scholar]
  13. Cavanna AE,Trimble MR ( 2006): The precuneus: A review of its functional anatomy and behavioural correlates. Brain 129: 564–583. [DOI] [PubMed] [Google Scholar]
  14. Chee MW,O'Craven KM,Bergida R,Rosen BR,Savoy RL ( 1999): Auditory and visual word processing studied with fMRI. Hum Brain Mapp 7: 15–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Chierchia G ( 1995): The variability of impersonal subjects In: Bach E, Jelinek E, Kratzer A, Partee BH, editors. Quantification in Natural Language. Dordrecht: Kluwer; pp 107–143. [Google Scholar]
  16. Chomsky N ( 1965): Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. [Google Scholar]
  17. Chomsky N ( 1981): Lectures on Government and Binding. Dordrecht: Foris. [Google Scholar]
  18. Chomsky N ( 1982): Some Concepts and Consequences of the Theory of Government and Binding. Cambridge, MA: MIT Press. [Google Scholar]
  19. Clahsen H,Featherston S ( 1999): Antecedent priming at trace positions: Evidence from German scrambling. J Psycholinguist Res 28: 415–437. [Google Scholar]
  20. Cummins S,Roberge Y ( 2004): Null objects in French and English In: Auger J, Clements C, Vance B, editors. Contemporary Approaches to Romance Linguistics. Amsterdam: John Benjamins; pp 121–138. [Google Scholar]
  21. Den Ouden DB,Fix SC,Parrish T,Thompson CK ( 2009): Argument structure effects in action verb naming in static and dynamic conditions. J Neurolinguist 22: 196–215. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Dowty D ( 1978): Governed transformations as lexical rules in a Montague grammar. Linguistic Inquiry 9: 393–426. [Google Scholar]
  23. Dowty D ( 1989): On the semantic content of the notion of “thematic role” In: Chierchia G, Partee B, Turner R, editors. Properties, Types, and Meanings, Vol. 2 Dordrecht: Kluwer; pp 69–129. [Google Scholar]
  24. Engelberg S ( 2002): Intransitive accomplishments and the lexicon: The role of implicit arguments, definiteness, and reflexivity in aspectual composition. J Seman 19: 369–416. [Google Scholar]
  25. Featherston S,Gross M,Münte TF,Clahsen H ( 2000): Brain potentials in the processing of complex sentences: An ERP study of control and raising constructions. J Psycholinguist Res 29: 141–154. [DOI] [PubMed] [Google Scholar]
  26. Fellbaum C ( 1999): The Organisation of Verbs and Verb Concepts in a Semantic Net. Dordrecht: Kluwer. [Google Scholar]
  27. Fiebach CJ,Schlesewsky M,Friederici AD ( 2001): Syntactic working memory and the establishment of filler‐gap dependencies: Insights from ERPs and fMRI. J Psycholinguist Res 30: 321–338. [DOI] [PubMed] [Google Scholar]
  28. Fillmore CJ ( 1986): Pragmatically controlled zero anaphora. Berkeley Linguist Soc 12: 95–107. [Google Scholar]
  29. Friedmann N,Gvion A ( 2007): As far as individuals with conduction aphasia understood these sentences were ungrammatical: Garden path in conduction aphasia. Aphasiology 21: 570–586. [Google Scholar]
  30. Friedmann N,Taranto G,Shapiro LP,Swinney D ( 2008): The leaf fell (the leaf): The online processing of unaccusatives. Linguist Inquiry 39: 355–377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Friston KJ,Holmes AP,Worsley KJ,Poline JB,Frith CD,Frackowiak RSJ ( 1995): Statistical parametric maps in functional imaging: A general linear approach. Hum Brain Mapp 2: 189–210. [Google Scholar]
  32. Frost R,Plaut DC ( 2005): The word‐frequency database for printed Hebrew.http://word-freq.mscc.huji.ac.il/index.htm.
  33. García Velasco D,Portero Muñoz C ( 2002): Understood objects in functional grammar. Working Papers in Functional Grammar. 76. Amsterdam: Functional Grammar Foundation. [Google Scholar]
  34. Grimshaw J ( 1979): Complement selection and the lexicon. Linguist Inquiry 10: 279–326. [Google Scholar]
  35. Groefsema M ( 1995): Understood arguments: A semantic/pragmatic approach. Lingua 96: 139–161. [Google Scholar]
  36. Grodzinsky Y ( 1995): Trace‐deletion, theta‐roles, and cognitive strategies. Brain Lang 51: 469–497. [DOI] [PubMed] [Google Scholar]
  37. Holmes VM ( 1987): Syntactic parsing: In search of the garden path In: Coltheart M, editor. Attention and Performance XII. Hillsdale, NJ: Erlbaum; pp 587–599. [Google Scholar]
  38. Kemeny S,Xu J,Park GH,Hosey LA,Wettig CM,Braun AR ( 2006): Temporal dissociation of early lexical access and articulation using a delayed naming task: An FMRI study. Cereb Cortex 16: 587–595. [DOI] [PubMed] [Google Scholar]
  39. Kluender R,Kutas M ( 1993): Bridging the gap: Evidence from ERPs on the processing of unbounded dependencies. J Cogn Neurosci 5: 196–214. [DOI] [PubMed] [Google Scholar]
  40. Lancaster JL,Summerin JL,Rainey L,Freitas CS,Fox PT ( 1997): The Talairach Daemon, a database server for Talairach Atlas Labels. NeuroImage 5: S633. [Google Scholar]
  41. Lasnik H ( 2002): The minimalist program in syntax. Trends Cogn Sci 6: 432–437. [DOI] [PubMed] [Google Scholar]
  42. Manzini MR ( 1992): The projection principle(s): A reexamination In: Roca IM, editor. Thematic Structure: Its Role in Grammar. Berlin: Foris; pp 271–291. [Google Scholar]
  43. Mummery CJ,Patterson K,Hodges JR,Price CJ ( 1998): Functional neuroanatomy of the semantic system: Divisible by what? J Cogn Neurosci 10: 766–777. [DOI] [PubMed] [Google Scholar]
  44. Nichols T,Brett M,Andersson J,Wager T,Poline JB ( 2005): Valid conjunction inference with the minimum statistic. NeuroImage 25: 653–660. [DOI] [PubMed] [Google Scholar]
  45. Nicol J,Swinney D ( 1989): The role of structure in conference assignment during sentence comprehension. J Psycholinguist Res 18: 5–19. [DOI] [PubMed] [Google Scholar]
  46. Nobre AC,Allison T,McCarthy G ( 1994): Word recognition in the human inferior temporal lobe. Nature 372: 260–263. [DOI] [PubMed] [Google Scholar]
  47. Oldfield RC ( 1971): The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia 9: 97–113. [DOI] [PubMed] [Google Scholar]
  48. Paus T,Koski L,Caramanos Z,Westbury C ( 1998): Regional differences in the effects of task difficulty and motor output on blood flow response in the human anterior cingulate cortex: A review of 107 PET activation studies. NeuroReport 9: R37–R47. [DOI] [PubMed] [Google Scholar]
  49. Poeppel D,Guillemin A,Thompson J,Fritz J,Bavelier D,Braun A ( 2004): Auditory lexical decision, categorical perception, and FM direction discrimination differentially engage left and right auditory cortex. Neuropsychologia 42: 183–200. [DOI] [PubMed] [Google Scholar]
  50. Reinhart T ( 2000): The Theta System: Syntactic Realization of Verbal Concepts. OTS Working Papers in Linguistics. Utrecht: Utrecht University, OTS. [Google Scholar]
  51. Resnik P ( 1996): Selectional constraints: An information‐theoretic model and its computational realization. Cognition 61: 127–159. [DOI] [PubMed] [Google Scholar]
  52. Rice S ( 1988): Unlikely lexical entries In: Jaisser SAA, Singmaster H, editors. Proceedings of the 14th Annual Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society. [Google Scholar]
  53. Rizzi L ( 1986): Null objects in Italian and the theory of pro. Linguist Inquiry 17: 501–557. [Google Scholar]
  54. Shapiro LP,Levine BA ( 1990): Verb processing during sentence comprehension in aphasia. Brain Lang 38: 21–47. [DOI] [PubMed] [Google Scholar]
  55. Shapiro LP,Zurif E,Grimshaw J ( 1987): Sentence processing and the mental representation of verbs. Cognition 27: 219–246. [DOI] [PubMed] [Google Scholar]
  56. Shapiro LP,Zurif E,Grimshaw J ( 1989): Verb processing during sentence comprehension: Contextual impenetrability. J Psycholinguist Res 18: 223–243. [DOI] [PubMed] [Google Scholar]
  57. Shapiro LP,Brookins B,Gordon B,Nagel N ( 1991): Verb effects during sentence processing. J Exp Psychol Learn Mem Cogn 17: 983–996. [DOI] [PubMed] [Google Scholar]
  58. Shapiro LP,Gordon B,Hack N,Killackey J ( 1993): Verb argument processing in complex sentences in Broca and Wernicke's aphasia. Brain Lang 45: 423–447. [DOI] [PubMed] [Google Scholar]
  59. Shetreet E,Palti D,Friedmann N,Hadar U ( 2007): Cortical representation of verb processing in sentence comprehension: Number of complements, subcategorization and thematic frames. Cereb Cortex 17: 1958–1969. [DOI] [PubMed] [Google Scholar]
  60. Shetreet E,Friedmann N,Hadar U ( 2009a): An fMRI study of syntactic layers: Sentential and lexical aspects of embedding. NeuroImage 48: 707–716. [DOI] [PubMed] [Google Scholar]
  61. Shetreet E,Friedmann N,Hadar U ( 2009b): The cortical representation of unaccusative verbs. Lang Brain 8: 21–29 (in Hebrew). [Google Scholar]
  62. Tanenhaus MK,Boland J,Garnsey SM,Carlson GN ( 1989): Lexical structure in parsing long‐distance dependencies. J Psycholinguist Res 18: 37–50. [DOI] [PubMed] [Google Scholar]
  63. Thompson SA,Hopper PJ ( 2001): Transitivity, clause structure and argument structure: Evidence from conversation In: Hopper PJ, editor. Frequency and the Emergence of Linguistic Structure. Amsterdam: John Benjamins; pp 27–60. [Google Scholar]
  64. Thompson CK,Bonakdarpour B,Fix SC,Blumenfeld HK,Parrish TB,Gitelman DR,Mesulam MM ( 2007): Neural correlates of verb argument structure processing. J Cogn Neurosci 19: 1753–1767. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Trueswell JC,Tanenhaus MK,Kello C ( 1993): Verb‐specific constraints in sentence processing: Separating effects of lexical preference from garden‐paths. J Exp Psychol Learn Mem Cogn 19: 528–553. [DOI] [PubMed] [Google Scholar]
  66. Vandenberghe R,Price C,Wise R,Josephs O,Frackowiak RSJ ( 1996): Functional anatomy of a common semantic system for words and pictures. Nature 383: 254–256. [DOI] [PubMed] [Google Scholar]
  67. Van Valin RD ( 2001): An Introduction to Syntax. Cambridge, UK: Cambridge University Press. [Google Scholar]
  68. Van Valin RD,LaPolla RJ ( 1997): Syntax: Structure, Meaning and Function. Cambridge, UK: Cambridge University Press. [Google Scholar]
  69. Wagner AD,Schacter DL,Rotte M,Koutstaal W,Maril A,Dale AM,Rosen BR,Buckner RL ( 1998): Building memories: Remembering and forgetting of verbal experiences as predicted by brain activity. Science 281: 1188–1191. [DOI] [PubMed] [Google Scholar]
  70. Warburton EK,Wise R,Price CJ,Weiller C,Hadar U,Ramsay S,Frackowiak RSJ ( 1996): Noun and verb retrieval by normal subjects: Studies with PET. Brain 119: 159–179. [DOI] [PubMed] [Google Scholar]
  71. Zurif EB,Swinney D,Prather P,Solomon J,Bushell C ( 1993): An on‐line analysis of syntactic processing in Broca's and Wernicke's aphasia. Brain Lang 45: 448–464. [DOI] [PubMed] [Google Scholar]

Articles from Human Brain Mapping are provided here courtesy of Wiley

RESOURCES