The impact of semantic reference on word class: An fMRI study of action and object naming
Introduction
Nouns and verbs are the basic building blocks of all known languages. All languages use some linguistic expressions to indicate “things”, however abstract and intangible they might be, and other expressions to identify parts of reality as “events”, or “actions” (Langacker, 1987, Tomasello, 1998, Tomasello et al., 1997). Typically, and this is certainly true of the European languages in which most of the investigations have been conducted, nouns and verbs have also very distinct morphological and syntactic properties (Druks, 2002).
There is a considerable body of neuropsychological and neuroimaging evidence supporting the distinction between the brain correlates of noun and verb processing. Crucial information comes from studies of brain-damaged patients. There is ample evidence that aphasic patients may be impaired in the processing of verbs while nouns are relatively spared and vice versa (Cappa et al., 1998, Daniele et al., 1994, Hillis et al., 2002, McCarthy and Warrington, 1985, Miceli et al., 1984). A difference in lesion localization has been suggested to underlie this behavioral dissociation, with noun processing disorders associated with lesions in the left temporal lobe, while patients impaired in verb processing typically have large lesions involving the left fronto-parietal cortex. This dissociation has often been interpreted as evidence that damage to distinct brain areas can selectively affect the ability to use nouns or verbs, suggesting that different neural networks subserve the two word classes (Cappa and Perani, 2003, Shapiro and Caramazza, 2003a, Shapiro and Caramazza, 2003b). A partial dissociation was reported by Tranel and colleagues: in a study conducted on a large group of patients with unilateral lesions, these authors found that lesions in left frontal and parietal regions were associated with deficits in naming both actions and objects, while lesions in anterior and inferior temporal regions lead to impairment in naming concrete objects, and not actions (Tranel et al., 2001).
However, the issue is complicated by the lack of consistent results from imaging investigations. Although some studies have revealed incompletely overlapping neural substrates for noun and verb processing, none has convincingly shown a neurological double dissociation (i.e., each word type being associated with the activation of at least partially independent brain systems). Evidence for a noun–verb difference comes from a PET experiment that tested lexical decision in healthy adults (Perani et al., 1999). The results indicated incompletely overlapping neural substrates for verb and noun processing. Verb-specific activations were reported in dorsolateral frontal, parietal and lateral temporal cortex, while no areas were activated specifically for nouns. In the same direction, recent evidence from event-related potential (ERP) studies found reliable differences in the ERP response evoked by verbs compared to nouns in the 300- to 800-ms time window in a semantic priming task (Khader et al., 2003). Verbs were associated to a larger positivity over central to frontal areas and a larger negativity over occipital and temporo-parietal regions. No differences were observed in the topography of the N-400 component. Khader and Rosler (2004) also report a larger power attenuation in the θ frequency band of the electroencephalogram at left anterior sites, accompanied by a decrease of coherence while subjects read phrases beginning with verbs compared to nouns. This finding was interpreted as reflecting word class differences. In a recent experiment, Tyler and colleagues reported a small verb-specific activation in left inferior frontal gyrus when regularly inflected nouns and verbs were contrasted, a result attributed to differences in the morphosyntactic characteristics of the two categories (Tyler et al., 2004). Other studies, however, failed to show any difference between closely matched nouns and verbs in lexical decision and semantic judgment, as well as during verb and noun generation tasks (Soros et al., 2003, Tyler et al., 2001, Warburton et al., 1996).
As mentioned above, nouns and verbs differ on numerous syntactic and morphological dimensions, but they also have different semantic content, with most verbs referring to actions or events and most nouns referring to objects. With a few exceptions (see, for example, Shapiro and Caramazza, 2003b, Tranel et al., 2003), patient studies have employed tasks that do not allow a separate consideration of the role of semantic and grammatical factors in producing the observed noun–verb difference. In the few cases where semantic factors have been examined, patient studies have suggested that specific semantic dimensions might modulate the neural representation of objects and actions. A recent lesion study by Tranel and colleagues showed a partial overlap between systems supporting conceptual knowledge for actions – performed with different body parts – and tools, while discrete systems appeared to underlie actions versus face representations. The authors suggest that overlapping visuomotor representations might be part of tool and action knowledge (Tranel et al., 2003).
The discrepancy between the relatively consistent results provided by the lesion literature and the contradictory imaging data might be a result of the differences in methodology, with imaging revealing larger sets of activated areas involved in system processing than the “necessary” areas isolated by lesion studies (Corina et al., 2005). However, the issue is certainly complicated by the interplay of semantic, morphological and syntactic factors, in addition to stimulus-specific characteristics, such as imageability, frequency and age of acquisition.
In most studies, efforts have been made to control for confounds known to influence processing, but, with very few exceptions (Oliveri et al., 2004, Perani et al., 1999, Tyler et al., 2003), no study has focused on semantic distinctions as factors of interest. The “semantic” distinction between nouns and verbs can be viewed at multiple levels. One dimension which has been suggested to play a significant role is the more “abstract” nature of verbs in comparison to nouns. This hypothesis, originally formulated by Allport and Funnell (1981), has been recently proposed in terms of differences in imageability. Other authors (Bird et al., 2000a, Bird et al., 2003) found that the verb deficit shown by a group of patient in confrontation naming tasks was actually due to the lower imageability of verbs. In a norming study, the same authors found that, in the case of noun/verb homonyms (the crack-to crack), the verb forms were rated as less imageable than the noun forms (Bird et al., 2000b). However, Berndt et al. (2002) paint a more complex picture, compatible with independent effects of imageability and grammatical class. In a picture-naming task, some aphasic patients found verbs more difficult to retrieve compared to nouns, even when imageability of the two classes was equated. Other patients were sensitive to imageability, irrespective of word class, while others showed effects of both imageability and word class. A significant link was found in their group of aphasic patients between sentence production impairment, rather than anomia, and defective verb retrieval. Silveri and Cappa (2003) propose that verbs depend less on semantic activation and more on the interaction between syntactic specification and the abstract phonological representation held in short-term memory. A different account in term of semantic features has been proposed by Breedin et al. (1998) on the basis of their findings in a group of aphasic patients with a selective verb retrieval disorder. These patients were found to be more impaired in the retrieval of semantically simple than semantically complex verbs (give versus donate). Among other likely explanations, the authors mention the possibility that verbs with more semantic features might have a richer and more stable representation, thus making complex verbs easier to retrieve. On the other hand, Collina et al. (2001) found that the number of arguments played a crucial role: the verb-specific impairment shown by an agrammatic patient was abolished when verbs and nouns were matched in terms of argument structure, with the patient becoming impaired for both verbs and nouns associated with multiple arguments.
The role of specific semantic factors has been less extensively investigated. Many verbs refer to actions and movement, and many nouns refer to objects. Indeed, the dynamic versus static distinction is central to the very notion of, respectively, verb and noun (Langacker, 1987). It is worth underlining that these more “typical” verbs and nouns are usually assessed in picture naming, which is the task most commonly used in testing aphasic patients. Beyond this basic difference, some particularly relevant semantic distinctions might be a dominant organizing principle within and across word classes. One such potential principle is the notion of manipulation and manipulability. There is ample evidence suggesting that the way we act on objects has an impact on how the information about those objects is represented and retrieved (Binkofski et al., 1999b, Chao and Martin, 2000, Grafton et al., 1997, Oliver and Thompson-Schill, 2003). In particular, activation of the same parietal premotor circuits involved in manipulation has been reported during the observation of manipulable objects (Grezes et al., 2003). According to this view, knowledge about object shape and size might be acquired by grasping, so that areas that mediate visually guided grasping (Murata et al., 1997, Murata et al., 1996, Murata et al., 2000, Sakata et al., 1999) become part of the object representation. Other evidence points to the involvement of visuomotor networks in action representation. Motor programs equivalent to those used during action execution are also involved in observing and understanding actions made by others (Rizzolatti et al., 2001). The neural circuits involved are somatotopically organized and effector specific (Buccino et al., 2001, Kohler et al., 2002, Pulvermuller, 2005) and are supposed to code actions at the abstract level of a shared motor representation. Recent neuroimaging studies have explored the neural network involved in motor execution, observation, imagination and verbalization, reporting that mirror neurons are engaged by action-related noises in monkeys (Kohler et al., 2002), and that in humans, fronto-parietal motor networks are active during the visual presentation of action words (Hauk et al., 2004), and the auditory presentation of action sentences, with hand-, foot-, and mouth-related actions eliciting somatotopically organized activations in motor and premotor cortex (Tettamanti et al., 2005). A meta-analysis by Grezes and Decety (2001) found that, while execution, observation and imagination were associated with largely overlapping activations, “action-related” verbalizations were associated with a partially distinct substrate, and consistently left-lateralized. This result however may be confounded by the fact that none of the studies included in the analysis required the production of verbs related to specific actions.
If the processing of action-related words activates the motor circuits involved in the execution of the same actions, and if motion-related information is part of object representations, we would expect reference to manipulation to modulate the activation for both verbs and nouns. Both nouns and verbs may be subclassified as related or unrelated to manipulation, based on the possibility of manipulating the object in the case of nouns, or the nature of the action in the case of verbs. The hypothesis can be entertained that this semantic distinction may modulate the word class dissociation (Gerlach et al., 2002, Kellenbach et al., 2003). Recent work by Olivieri et al. brings support to this view. These authors applied transcranial magnetic stimulation (TMS) to the left motor cortex of subjects during processing of action-related verbs and nouns (action verbs mostly referred to manipulations, and all action nouns referred to manipulable objects), reporting a modulation of excitability of motor cortex that was not observed during the processing of abstract/psychological verbs and non-manipulable nouns. Producing an action-related word lead to an increase in neural activity in the left motor cortex, regardless of the grammatical category of the word produced (Oliveri et al., 2004).
In order to test impact of manipulability on the processing of nouns and verbs, we used event-related fMRI to contrast brain activity while naming pictures representing artifacts that can or cannot be manipulated and actions that involve fine hand movements versus whole-body actions. In particular, we expected hand-related items to elicit activations in motor and premotor areas, irrespective of word class. Based on the evidence from Oliveri et al. (2004), among others, we expected that differences between word classes, when the main semantic dimension was equated, should be minimal. A picture-naming task was chosen for a number of reasons. Picture naming is a widely used technique for the study of lexical access and real-time language processing (for reviews, see Glaser, 1992, Johnson et al., 1996, Murtha et al., 1999, Snodgrass and Vanderwart, 1980). As a measure of lexical access, it allows the investigation of fundamental components of conceptual organization and language processing (Buckner et al., 1995, Buckner et al., 1996, McCarthy et al., 1993, Ojemann et al., 1998, Petersen et al., 1988, Petersen et al., 1998, Posner et al., 1988). It also has a strong clinical relevance: word-finding deficits (anomia) are observed in every form of adult aphasia (Goodglass, 1993), in the elderly (Hodgson and Ellis, 1998), and in developmental language impairment (Lahey and Edwards, 1999).
Section snippets
Participants
Thirteen right-handed (Oldfield, 1971) volunteers participated in the study (4 females and 9 males; mean age 23.4 years, range 21 to 26 years). All participants had a comparable education level, were native speakers of Italian with little or no exposure to other languages before the age of 11, and had no history of learning disabilities or psychiatric and neurological disorders. All had normal hearing and normal or corrected-to-normal vision. Participants provided written consent to
Behavioral results during fMRI data acquisition
Subjects produced an acceptable response on 85% (SE = 3.5) of action trials and 92% (SE = 2.3) of noun trials. Responses were considered acceptable if they matched the dominant responses produced by Italian subjects in the IPNP corpus norms, including alterations such as diminutives (“Italian:barchetta/little boat” when the target was “It.barca/boat”), reductions (“It.ferro/iron” when the target was “It.ferro da stiro/tool for ironing”), or expansions (“It.molletta per i panni/clothespin” when
Discussion
The main finding of this study is that, for a picture naming task, a specific sensorimotor dimension, i.e., manipulability, played a major role in modulating the pattern of brain activation during noun and verb processing. In contrast, we observed that the word-class (noun–verb) distinction did not result in significant differences in activation. Further exploration of the data at a more liberal significance threshold revealed a limited set of differences in the pattern of brain activity
Acknowledgments
We thank Davide Anchisi and Marco Tettamanti for their technical expertise. This research was supported by Human Frontier Science Program Grant RG148/2000, by NIH grant 3R01DC00216-21 and by the 5th European Program for the project “Connectivity in language rehabilitation” in Europe (LSDE).
References (110)
- et al.
The inferential impact of global signal covariates in functional neuroimaging analyses
NeuroImage
(1998) - et al.
A global estimator unbiased by local changes
NeuroImage
(2001) - et al.
Grammatical class and imageability in aphasic word production: their effects are independent
J. Neurolinguist.
(2002) - et al.
Why is a verb like an inanimate object? Grammatical category and semantic category deficits
Brain Lang.
(2000) - et al.
The rise and fall of frequency and imageability: noun and verb production in semantic dementia
Brain Lang.
(2000) - et al.
Verbs and nouns: the importance of being imageable
J. Neurolinguist.
(2003) - et al.
Semantic factors in verb retrieval: an effect of complexity
Brain Lang.
(1998) - et al.
The neural correlates of noun and verb processing
J. Neurolinguist.
(2003) - et al.
Representation of manipulable man-made objects in the dorsal stream
NeuroImage
(2000) - et al.
The role of argument structure in the production of nouns and verbs
Neuropsychologia
(2001)
Evidence for a possible neuroanatomical basis for lexical processing of nouns and verbs
Neuropsychologia
Removal of confounding effects of global signal in functional MRI analyses
NeuroImage
Nouns and verbs: a review of the literature
J. Neurolinguist.
Characterizing dynamic brain responses with fMRI: a multivariate approach
NeuroImage
Event-related fMRI: characterizing differential responses
NeuroImage
Thresholding of statistical maps in functional neuroimaging using the false discovery rate
NeuroImage
Picture naming
Cognition
fMRI BOLD response to increasing task difficulty during successful paired associates learning
NeuroImage
Premotor cortex activation during observation and naming of familiar tools
NeuroImage
Perception of coherent motion, biological motion and form-from-motion under dim-light conditions
Vision Res.
Brain activity evoked by inverted and imagined biological motion
Vision Res.
Brain areas active during visual perception of biological motion
Neuron
The human action recognition system and its relationship to Broca's area: an fMRI study
NeuroImage
Somatotopic representation of action words in human motor and premotor cortex
Neuron
Last in, first to go: age of acquisition and naming in the elderly
Brain Lang.
EEG power and coherence analysis of visually presented nouns and verbs reveals left frontal processing differences
Neurosci. Lett.
Differences between noun and verb processing in a minimal phrase context: a semantic priming study using event-related brain potentials
Brain Res. Cogn. Brain Res.
Perceptual specificity in visual object priming: functional magnetic resonance imaging evidence for a laterality difference in fusiform cortex
Neuropsychologia
fMRI of healthy older adults during Stroop interference
NeuroImage
Visual search and memory search engage extensive overlapping cerebral cortices: an fMRI study
NeuroImage
Semantic memory and the brain: structure and processes
Curr. Opin. Neurobiol.
Category specificity in an agrammatic patient: the relative impairment of verb retrieval and comprehension
Neuropsychologia
On the basis for the agrammatic's difficulty in producing main verbs
Cortex
The assessment and analysis of handedness: the Edinburgh inventory
Neuropsychologia
Anatomy and time course of discrimination and categorization processes in vision: an fMRI study
NeuroImage
Localization of body parts in brain injured subjects
Neuropsychologia
Grammatical processing of nouns and verbs in left frontal cortex?
Neuropsychologia
The representation of grammatical categories in the brain
Trends Cogn. Sci.
Segregation of the neural correlates of language and phonological short-term memory
Cortex
Naming actions and objects: cortical dynamics in healthy adults and in an anomic patient with a dissociation in action/object naming
NeuroImage
Components of the mental lexicon
Trans. R. Soc. London
Introducing the CRL International Picture-Naming Project (CRL-IPNP)
Center Res. Lang. Newsl.
Timed picture naming in seven languages
Psychon. Bull. Rev.
fMRI responses to video and point-light displays of moving humans and manipulable objects
J. Cogn. Neurosci.
A fronto-parietal circuit for object manipulation in man: evidence from an fMRI-study
Eur. J. Neurosci.
A parieto-premotor network for object manipulation: evidence from neuroimaging
Exp. Brain Res.
Neural correlates of mental transformations of the body-in-space
Proc. Natl. Acad. Sci. U. S. A.
Neural systems for tactual memories
J. Neurophysiol.
Specific involvement of human parietal systems and the amygdala in the perception of biological motion
J. Neurosci.
Action observation activates premotor and parietal areas in a somatotopic manner: an fMRI study
Eur. J. Neurosci.
Cited by (103)
Neural correlates for nouns and verbs in phrases during syntactic and semantic processing: An fMRI study
2020, Journal of NeurolinguisticsThe motor features of action verbs: fMRI evidence using picture naming
2018, Brain and LanguageCitation Excerpt :Given the noun-verb dissociation discussed in the literature is mostly based on picture naming performance (for reviews see Cotelli et al., in press; Mätzig et al., 2009; Vigliocco et al., 2011), the current study focused mainly on picture naming. Within this context, it is surprising that the motor features of action verbs have so far been little illustrated in imaging studies which addressed the noun-verb dissociation in neurologically intact participants using picture naming (Berlingeri et al., 2008; Damasio et al., 2001; Liljeström et al., 2008; Momenian et al., 2016; Saccuman et al., 2006; Siri et al., 2008; Tranel et al., 2005). Rather, those studies have yielded mixed results which can be attributed to, except for morpho-syntactic factors, various unmatched psycholinguistic variables.