Where in the brain is nonliteral language? A coordinate-based meta-analysis of functional magnetic resonance imaging studies
Highlights
► Coordinate-based meta-analysis of nonliteral language fMRI studies. ► Analysis includes research on metaphors, metonymy, idioms and irony. ► A predominantly left lateralised fronto-temporal network appears crucial. ► The right hemisphere plays a smaller, but significant role for nonliteral stimuli. ► Both hemispheres contribute to processing of novel stimuli.
Introduction
‘Neuroimaging is a gold mine’, is a phrase that the reader quickly understands is intended as a metaphor, not a literal statement (Glucksberg, 2003). Everyday language is remarkably rich in such non-literal expressions (Gedigan et al., 2006, Gibbs, 2000, Lakoff and Johnson, 1980, Markert and Hahn, 2002, Pollio et al., 1977, Whalen et al., 2009). Taken literally or ‘word by word’, non-literal expressions are either incorrect entirely or incorrect in the particular context given. Metaphors, proverbs, idioms, metonymy, and ironic or sarcastic expressions are among the most important forms of non-literal language. Verbal metaphors, such as the example above, bring together semantic entities that are ‘normally’ not related and are used for conceptualising and expressing relevant parts of our lives that are otherwise difficult to explain. In the case of ironic remarks, (for example, stating ‘oh brilliant’ when something bad happens), in many cases the speaker literally says the exact opposite of what is intended. Proverbs are popularly known sayings based on common sense or practical experience. Although they often are metaphorical, they do not have to be. Like idioms, they are structurally ‘frozen’, meaning they are fixed expressions. Idioms form a large group in terms of syntactic and semantic characteristics, although like all non-literal expressions, idioms can be short and of simple syntax. Some idioms, much like metaphors, can be understood by comparing the semantic entities within, whereas others cannot (for example, ‘kick the bucket’, which in English is an idiomatic expression for ‘to die’).
Scientific discussion about the cognitive operations behind understanding correct non-literal meanings dates back to Aristotle (Giora, 2007). It is an ongoing debate whether non-literal expression requires unique cognitive processes (Giora, 1997). However, all types of non-literal expressions have in common that the recipient must realise that the statement is intended non-literally, not literally. Some researchers assume the literal meaning is always processed first (Ortony, 1993), subsequently rejected, reanalysed, and replaced by the correct non-literal meaning. This assumption is compatible with a role for the inferior frontal gyrus in non-literal language comprehension, which plays a role in detecting semantic violations, reanalysing meanings, and selection among competing alternatives. It was hypothesised that the medial prefrontal cortex could play role in suppression of alternative literal meanings during non-literal language comprehension (Papagno and Romero-Lauro, 2010). However, several lines of evidence indicate that in the proper context, non-literal expressions are processed with equal speed (Glucksberg, 2003), indicating that there is no initial dominance of literal meanings. Beeman (Jung-Beeman, 2005) proposed the right cerebral hemisphere has a key role in processing non-literal expressions. His ‘coarse semantic coding theory’ (Beeman, 1998, Beeman et al., 1994) proposes that the left hemisphere (LH) specialises in analysing fine, or close, semantic relationships but the right hemisphere (RH) is specialised in both fine and coarse, or distant, semantic relationships. Mapping distant semantic fields during non-literal expressions would result in RH brain activity. According to this theory, metaphoric and other non-literal relationships represent distant relationships. Applied to non-literal language, Beeman’s theory thus suggests that RH activation during non-literal language is crucial and thereby consistently observed. However, the results on lateralisation non-literal language are mixed in terms of RH contribution (Anaki et al., 1998, Giora, 2007, Rapp et al., 2007, Winner and Gardner, 1977).
In addition, other non-literal language cognitive operations have attracted research attention. Particularly ironic remarks require perspective taking and theory of mind processes (Colston and Gibbs, 2002, Happé, 1996) so irony comprehension tasks were used as paradigms to investigate theory of mind in clinical populations (Sprong et al., 2007). Non-literal language processing may require an increased demand for context and general world knowledge integration (Giora, 2003, Glucksberg, 2003, Rapp et al., 2011). The figurativeness of non-literal language inspired researchers to use metaphors (Aziz-Zadeh et al., 2006) and idioms (Boulenger et al., 2009) as stimuli to investigate motor semantics. The ability to comprehend non-literal language is impaired in a number of clinical populations (Gernsbacher and Pripas-Kapit, 2012, Thoma and Daum, 2006), including patients with schizophrenia (Rapp, 2009, Rapp and Schmierer, 2010), autistic subjects (Martin and McDonald, 2004), and patients with dementia (Rapp and Wild, 2011) or other neurodevelopmental disorders (Annaz et al., 2009).
During the last decade, knowledge regarding the functional neuroanatomy of non-literal language has increased significantly with the publication of more than 50 additional studies. In addition to new studies with brain-lesioned patients (Bélanger et al., 2009, Brownell et al., 2007, Cacciari et al., 2006, Lundgren et al., 2011, Papagno and Caporali, 2007, Papagno et al., 2004, Papagno et al., 2006, Paul et al., 2003, Rinaldi et al., 2004, Shamay-Tsoory and Aharon-Peretz, 2007, Shamay-Tsoory et al., 2005, Zaidel et al., 2002), a number of studies used functional magnetic resonance imaging (fMRI) to investigate the neural correlates of non-literal language. In the first non-literal language fMRI study, Rapp et al. (2004) investigated syntactically simple metaphors matched for tense and word frequency, such ‘the lover´s words are harp sounds’, compared to literal control stimuli, such as ‘the lover´s words are lies’. Both hemispheres were involved in the metaphoric and literal stimuli contrasts versus baseline. Contrary to the study hypothesis, no RH activation was detected in the direct comparison. Metaphoric stimuli, relative to matched literal stimuli, activated a left fronto-temporal network with maxima in the anterior–inferior part of the left inferior frontal gyrus (IFG, Brodmann area (BA) 45/47), anterior temporal (BA 20), and posterior middle/inferior temporal (BA 37) gyri. These data were divergent from a previous positron emission tomography (PET) study by Bottini et al. (1994), in which RH activation was observed for phrasal metaphoric expressions. Subsequently, more than 15 additional fMRI metaphor studies followed (Table 1). The first fMRI studies on ironic expressions were published in 2006 (Eviatar and Just, 2006, Uchiyama et al., 2006, Wang et al., 2006a), followed by work on idioms (Zempleni, 2006) and metonymy (Rapp et al., 2011). In their first study, Rapp and colleagues determined that the anterior–inferior part of the left IFG in particular plays a key role in disentangling non-literal meanings (Kircher et al., 2007, Rapp et al., 2004). It remains unknown, however, if this prediction can be verified when all available evidence is merged in meta-analyses.
Other research questions on the functional anatomy of non-literal language remain unclear as well. The most discussed aspect in fMRI studies on non-literal language is the contribution of the right cerebral hemisphere. A majority of metaphor fMRI studies aimed to demonstrate a significant contribution of the right cerebral hemisphere. Narrative reviews indicate RH contribution is indeed present in some (Schmidt et al., 2010), but not all (Rapp et al., 2007), fMRI studies. However, no study has yet quantified the RH contribution by meta-analyses or analyses of the number of reported activation maxima.
A widely used but relatively loosely defined subdivision is the distinction between salient (sometimes called ‘fossilised’) and non-salient (or ‘novel’) non-literal expressions. Salient non-literal expressions are frequently used in everyday language, whereas a novel expression is not. It is widely accepted in the scientific literature that the salient and non-salient cognitive processes are different (Giora, 2003, Giora and Fein, 1999, Glucksberg, 2003) and this difference elicited interest in the literature regarding the neural correlates of non-literal language (Desai et al., 2011, Giora, 2007, Giora et al., 2000). The graded salience hypothesis (Giora, 1997), a linguistic theory that has received much attention in work on the neural correlates of metaphor and other non-literal stimuli, postulates a difference in comprehension processes between salient and non-salient non-literal expressions. A version of this hypothesis predicts that novel, but not salient, non-literal stimuli are predominantly processed by the right cerebral hemisphere (Giora, 2007, Giora et al., 2000). Although a number of fMRI studies aimed to test this hypothesis (Ahrens et al., 2007, Mashal et al., 2005, Mashal et al., 2007, Schmidt and Seger, 2009, Yang et al., 2009), there is still no consensus in the literature about the results.
In this work, we sought to merge findings in a coordinate based meta-analysis. We calculated activation-likelihood estimations (ALE) for non-literal language (NL language) fMRI studies. We included fMRI studies on metaphors, metonymy, idioms, proverbs, and ironic expressions. The ALE technique is a quantitative method for voxel-wise meta-analysis of neuroimaging or fMRI foci (Eickhoff et al., 2009, Turkeltaub et al., 2002). The ALE technique models the localisation uncertainty of reported neuroimaging foci as a three-dimensional Gaussian probability density distribution. The result is an estimation of ‘the likelihood that at least one of the foci in the literature should have been reported at this voxel’ (Turkeltaub and Coslett, 2010).
The main goal of this work was to compare the quantitative right and LH cluster contribution to non-literal language comprehension. Therefore, we analysed the available literature using different approaches: (I) compare the number of reported activation maxima in the hemispheres, and (II) compare the number of clusters in coordinate-based meta-analyses. We expected to find more LH clusters and activation maxima. We further expected that the results would support the graded salience hypothesis (Giora, 1997), that is, we expected a higher proportion of voxels in the RH for novel stimuli.
Section snippets
Materials and methods
Initially, fMRI studies on non-literal language were identified by literature search. Studies were identified with PUBMED and PSYCINFO databases using the search terms metaphor, proverb, non-literal, figurative, idiom, metonymy, irony, and sarcasm in combination with fMRI. Other studies the authors were familiar with were also included. Inclusion criteria were that the imaging modality was fMRI, the whole-brain analyses were from healthy, adult subjects, and differential contrasts for
Results
Our literature search identified 38 fMRI studies on non-literal language. Table 1 provides an overview of the literature. Seven fMRI studies on non-literal language were not included in the meta-analyses for the following reasons: (1) no coordinates reported in one study that was published within a book-chapter (Yu et al., 2006), (2) study only reports region of interest (ROI) analysis (Eviatar and Just, 2006, Mashal et al., 2005, Wang et al., 2006a), (3) no differential contrast for
Discussion
The aim of this work was to identify and meta-analyse all currently available fMRI studies on non-literal language. Non-literal language was defined as metaphor, metonymy, proverbs, idioms, or ironic/sarcastic remarks. The available studies cover a wide range of non-literal language (Table 1). Collectively, the technical quality of the available studies is satisfactory, with an average of 15 subjects and 43 stimuli.
The main conclusion from these meta-analyses is that a predominantly left
References (163)
- et al.
Functional MRI of conventional and anomalous metaphors in Mandarin Chinese
Brain Lang.
(2007) - et al.
Meeting of minds: the medial frontal cortex and social cognition
Nat. Rev. Neurosci.
(2006) - et al.
Broca's region: novel organizational principles and multiple receptor mapping
PLoS Biol.
(2010) - et al.
Cerebral hemispheric asymmetries in processing lexical metaphors
Neuropsychologia
(1998) - et al.
Comprehension of metaphor and metonymy in children with Williams syndrome
Int. J. Lang. Commun. Disord.
(2009) - et al.
Big words, halved brains and small worlds: complex brain networks of figurative language comprehension
PLoS One
(2011) - et al.
Congruent embodied representations for visually presented actions and linguistic phrases describing actions
Curr. Biol.
(2006) - et al.
Left ventrolateral prefrontal cortex and the cognitive control of memory
Neuropsychologia
(2007) - et al.
Perisylvian functional connectivity during processing of sentential negation
Front. Psychol.
(2011) - et al.
Decomposing metaphor processing at the cognitive and neural level through functional magnetic resonance imaging
Brain Res. Bull.
(2011)
Coarse semantic coding and discourse comprehension
Summation priming and coarse semantic coding in the right hemisphere
J. Cogn. Neurosci.
Use of prosodic cues in the production of idiomatic and literal sentences by individuals with right- and left-hemisphere damage
Brain Lang.
Differential behavioral effects in frontal lobe disease
Neuropsychologia
Where is the semantic system? A critical review and meta-analysis of 120 functional neuroimaging studies
Cereb. Cortex
Erp correlates of individual differences in the comprehension of nonliteral language
Metaphor. Symb.
Functional MRI of language: new approaches to understanding the cortical organization of semantic processing
Annu. Rev. Neurosci.
The role of the right hemisphere in the interpretation of figurative aspects of language. A positron emission tomography activation study
Brain
Grasping ideas with the motor system: semantic somatotopy in idiom comprehension
Cereb. Cortex
Appreciation of metaphoric alternative word meanings by left and right brain-damaged patients
Neuropsychologia
Assessing quality of metaphor interpretation by right hemisphere damaged patients
Brain Lang.
A system for scoring proverb interpretations provided by non-brain-damaged adults and aphasic adults
Clin. Aphasiol.
Mirror neurons and the understanding of behavioural symptoms in psychiatric disorders
Curr. Opin. Psychiatry
Neurocognitive mechanisms underlying metaphor comprehension and other figurative language
Metaphor Symb. Act.
Neuropsychological and neurophysiological correlates of idiom understanding: How many hemispheres are involved?
The comprehension of ambigious idioms in aphasic patients
Neuropsychologia
Literal, fictive and metaphorical motion sentences preserve the motion component of the verb: a TMS study
Brain Lang.
Interpretation of mentalistic actions and sarcastic remarks: effects of frontal and posterior lesions on mentalising
Neuropsychologia
Proverb interpretation in fluent aphasia and Alzheimer's disease: implications beyond abstract thinking
Aphasiology
Functional–anatomical organization of predicate metaphor processing
Brain Lang.
Atypical neural processing of ironic and sincere remarks in children and adolescents with autism spectrum disorders
Metaphor. Symb.
Are irony and metaphor understood differently?
Metaphor. Symb.
Higher-level language deficits resulting from left primary cerebellar lesions
Aphasiology
Theories of adults’ understanding and use of irony and sarcasm: applications to and evidence from research with children
Dev. Rev.
The neural career of sensory-motor metaphors
J. Cogn. Neurosci.
The influence of context on hemispheric recruitment during metaphor processing
J. Cogn. Neurosci.
The influence of sentence novelty and figurativeness on brain activity
Neuropsychologia
Functional near-infrared spectroscopy assessment of speech-related tasks
Brain Lang.
Lesion analysis of the brain areas involved in language comprehension
Cognition
Cortical correlates of auditory sensory gating: a simultaneous near-infrared spectroscopy (NIRS)—event-related potential (ERP) study
Neuroscience
Coordinate-based activation likelihood estimation meta-analysis of neuroimaging data: a random-effects approach based on empirical estimates of spatial uncertainty
Hum. Brain Mapp.
Brain correlates of discourse processing: an fMRI investigation of irony and conventional metaphor comprehension
Neuropsychologia
Understanding metaphoric sentences in the two cerebral hemispheres
Brain Cogn.
The extended language network: a meta-analysis of neuroimaging studies on text comprehension
Hum. Brain Mapp.
Further development in social reasoning revealed in discourse irony understanding
Child Dev.
Maturation of the language network: from inter- to intrahemispheric connectivities
PLoS One
Mechanisms of social cognition
Annu. Rev. Psychol.
The role of left prefrontal cortex in language and memory
Proc. Natl. Acad. Sci. U. S. A.
Processing of metaphoric and non-metaphoric alternative meanings of words after right- and left-hemispheric lesion
Brain Lang.
Cited by (166)
Proverb comprehension in schizophrenia: A comprehensive review and meta-analysis
2024, Comprehensive PsychiatryRight hemisphere and metaphor comprehension: A connectionist perspective
2023, NeuropsychologiaHemispheric dominance of metaphor processing for Chinese-English bilinguals: DVF and ERPs evidence
2022, Journal of Neurolinguistics