Although the core language areas of the brain are considered to be located in the left hemisphere (LH), there is robust evidence of the right hemisphere’s (RH) involvement in language processing at different levels. For instance, voice perception and phoneme processing are known to be bilaterally controlled (Obleser et al., 2008, Gupta and Padma Srivastava, 2020). Superior temporal gyri of both left and right hemispheres play a role in prelexical stages of speech comprehension (see, e.g., review by Price, 2010). A recent electrocorticography (ECoG) study showed the existence of a bilateral sublexical speech system, supported by neural circuits distributed across inferior-frontal, parietal, superior-temporal, premotor, and somatosensory cortices in both hemispheres, and activated during speech perception and production in an overt word-repetition task (Cogan et al., 2014). RH has also been shown to be involved in understanding natural language in context (such as text or conversations) and to contribute to semantic processing (Jung-Beeman, 2005). fMRI research showed that RH has a particular role in processing novel metaphors (in comparison with conventional ones), which was reflected by strong activity in the right homologue of Wernicke's area, right inferior-frontal gyrus and right posterior superior-temporal sulcus (Mashal et al., 2005, Mashal et al., 2007). ERP (event-related potential) studies also indicated that the processing of unpredictable (Federmeier et al., 2008) and relatively more novel (Coulson & Severens, 2007) semantic relations may be associated with RH. Studies using both fMRI and MEG have shown RH’s involvement in processing pragmatic (e.g., speech act-specific) information in a conversational context (Egorova et al., 2014, Egorova et al., 2016). The right hemisphere has also been shown to be involved in the relationship between spoken language and gesture in a study of communicative pragmatic intentions of naming and requesting (Tomasello et al., 2019), which suggested that word-gesture combinations speed up the neural comprehension processes. With respect to RH’s role in prosody processing, Sammler and colleagues (2015), using multimodal neuroimaging and brain stimulation, showed that it involves both dorsal and ventral pathways: for instance, when the right premotor cortex, which plays a crucial role in the dorsal stream, was subjected to inhibitory stimulation, it decreased participants' ability to categorise prosody. Furthermore, right-hemispheric homologues of core language areas are involved in language production, as shown using fMRI (van Ettinger-Veenstra et al., 2010). The right hemisphere has also been shown to contribute to grammatical processing, which has traditionally been attributed to core language areas of the LH. For instance, using fMRI, Bozic et al. (2015) demonstrated that, while complex grammatical computations are indeed predominantly supported by left‐lateralised frontotemporal networks, simple linear grammatical parsing is underpinned by temporal regions of both left and right hemispheres.

The RH may also play an important role in language processing when the left hemisphere is impaired. In the case of aphasia studies, it was found that bigger language network lesions in LH lead to more profound involvement of the right hemisphere in language processing (Barbieri et al., 2019), likely contributing to compensatory processes. There is evidence of changes in connectivity between language areas of LH and their right-hemispheric homologues in treatment-induced aphasia recovery (Vitali et al., 2010, Kiran et al., 2015). In an extreme case, when the entire LH is removed early in life, the language function can be fully supported by the RH alone with high efficiency (Ogden, 1996). On the other hand, RH damage may also lead to difficulties in language processing, for instance, in the comprehension of nonliteral language such as jokes (Brownell et al., 1983) and idioms (Van Lancker and Kempler, 1987) as well as intonation processing (Heilman et al., 1984). Such data allow concluding that recruitment of neural substrates in both right and left hemispheres may be instrumental in healthy language processing as well as in language recovery and amelioration in various deficits (for more information, see Dressel et al., 2010, Jokel et al., 2016, Meyer et al., 2017, Kiran and Thompson, 2019).

In sum, RH is involved in the language function at different processing levels, although its exact contributions are not clear. However, most neuroimaging studies so far have focussed on correlational evidence, which can show brain activation patterns in a language task but not the causal links between the areas activated and the function under consideration. To obtain causal evidence in a healthy brain, brain stimulation methods are required, such as transcranial magnetic stimulation (TMS) or transcranial direct current stimulation (tDCS). These techniques interfere with the neuronal functionality, leading to changes at the behavioural level when a functionally relevant structure is stimulated. A small number of TMS studies have addressed the specific contribution of RH Wernicke's area homologue (rW) to language comprehension, focussing mostly on semantic processing. For instance, online TMS during a verbal category membership task revealed that rW is particularly involved in semantic categorisation of atypical exemplars (those more distant from the category; Passeri et al., 2015). It was also shown that there is a difference between Wernicke's area and its right-hemispheric counterpart in processing dominant and subordinate meanings of ambiguous words (Harpaz et al., 2009). In that study, participants had to decide in a semantic decision task whether ambiguous word (e.g. ‘‘bank’’), which has both dominant ("a financial instittion") and subordinate ("river bank") meanings, was related or not to a subsequent word (that could have been related or not to the one of these meanings). During this task, TMS over CP5 or CP6 (depending on the session) was used twice – once in a dominant meaning block and once in a subordinate meaning block. CP5 and CP6 scalp positions are believed to be located above the Wernicke’s area and its right-hemisperic homologue, respectively (Fiori et al., 2011, Sparing et al., 2001). As a result, LH was shown to be more responsible for the dominant meanings, and RH - for the subordinate ones. Similar tDCS research, in turn, demonstrated that reactions to subordinate, but not dominant, associations were expedited by anodal stimulation of rW (as opposed to placebo/sham condition), while anodal tDCS of Wernicke’s area led to faster responses for dominant associations (Peretz & Lavidor, 2013).

It should be noted that the exact cortical geometry of Wernicke's area is still under discussion (Binder, 2015, Hertrich et al., 2020), which could be explained by differences between functional and neuroanatomical approaches to its definition (Karbe, 2014). For instance, Penfield and Roberts (1959) described the posterior part of the left middle temporal gyrus and the inferior parietal cortex as a «posterior speech cortex (Wernicke)». Some works defined Wernicke's area in the left posterior superior temporal gyrus (pSTG; Benson, 1979, DeWitt and Rauschecker, 2013), whereas other research additionally supports the idea of Wernicke's area extension toward inferior temporal, anterior temporal, and temporoparietal regions (BA 20, 37, 38, 39, 40; Ardila et al., 2016). Regardless of such differences, it is generally accepted that Wernicke’s area is centered around the left temporo-parietal junction (TPJ), with subregions in supramarginal gyrus, posterior STG, angular gyrus, and posterior superior temporal sulcus (STS), all of which are highly relevant for language comprehension and are well-connected to multiple frontal, temporal and other areas (Igelstrom et al., 2015). Correspondingly, Wernicke's area RH homologue is located in the vicinity of the right TPJ. It has been shown that bilateral TPJ contributes to pragmatic processing (Tomasello, 2023). For instance, it was demonstrated to be involved in processing indirectness in social communication (the listener's ability to infer the speaker's intended meaning relying on contextual cues), which may be related to the affective aspects of language use (Bašnáková et al., 2015). Interestingly, this activation was observed both when individuals participated in a conversation and when they only overheard parts of other conversations. TPJ is also associated with emotion perception and Theory of Mind (ToM) – the ability to infer and predict behavioural reactions, beliefs, thoughts, intentions, and plans of other people (for more details, see Saxe and Kanwisher, 2003; Mitchell & Phillips, 2015).

One particularly interesting question that involves bilateral contributions to language is related to the processing of abstract and concrete semantics. Conventionally, concrete words are considered to indicate material/physical objects, actions, or phenomena (e.g., table, walk, sunshine). In contrast, abstract ones refer to phenomena that have less clear physical references (Borghi & Binkofski, 2014) such as emotions (e.g., embarrassment), mental states (confusion), situations (war), and conditions (difficulty). The diversity of abstract concepts (emotions, numbers, evaluative concepts like aesthetic and moral ones) suggests not only different dimensions (sensorimotor, interoceptive, linguistic, social; Borghi et al., 2022) but also different neural representations (Borghi et al., 2018) for them. For instance, Desai with colleagues (2018), using neuroimaging meta-analyses, showed that the processing of numerical concepts, emotional concepts as well as moral ones is underpinned by the activation of different brain areas both in the right and left hemispheres. In particular, the above affective, emotional and social cognitive roles ascribed to the right TPJ suggest its involvement in abstract language.

That said, the special role of the right hemisphere in concrete vs. abstract word processing has been mostly demonstrated by clinical studies. Split-brain research demonstrated that isolated RH has poor ability in the comprehension of abstract words and better ability in recognising concrete ones (Zaidel & Schweiger, 1984). Furthermore, patients with global alexia, which emerged as a result of LH damage, demonstrated better reactions to highly imageable words in comparison with abstract and low imageable ones, also suggesting a relatively higher contribution of RH to concrete semantics (Larsen et al., 2004). The advantage of LH over RH for abstract vs. concrete words was found in healthy participants (Day, 1977, Deloche et al., 1987, Mohr et al., 1994), also suggesting a lack of support for abstract word representations in the RH (Lindell, 2006). For instance, the comparison of response times in the lexical decision task, in which words were presented in the left or right visual field (associated with the right and the left hemispheres, respectively), showed that the RH detected semantic relationships between concrete nouns and their superordinate categories, whereas abstract nouns were better recognized by the LH (Day, 1977). This evidence is in line with a theoretical framework claiming stronger left-hemispheric lateralisation of the cell assemblies (CAs) representing abstract words (which, for the lack of clear semantic references, must rely on core language areas in the LH), whereas the networks encoding concrete words may be associated with both hemispheres, depending on the exact sensorimotor reference and the corresponding modality areas involved (Pulvermüller, 1999). The term “cell assembly” refers to a neural circuit, an organised set of neurons, which has been formed through associative learning (Hebb, 1949). CAs can be formed due to repeated presentation of stimuli through co-activating different (visual, auditory, motor) areas (Constant et al., 2023), as confirmed by both neuroimaging and simulation studies, and can thus encode words, mental images and concepts, serving as their neural representations, which activate in response to a respective stimulus, allowing the individual to identify it (Huyck & Passmore, 2013). Acquiring semantic relationships between wordforms and their referents leads to the formation of distributed neuronal circuits connecting together both modality-specific (i.e., sensory, motor) cortical systems and core language areas serving as ‘lexico-semantic hubs’ in such networks (Tomasello et al., 2017). Naturally, such networks span across both hemispheres, incorporating different areas in LH and RH depending on the specific semantics.

Coming back to the use of non-invasive brain stimulation in this field, the acquisition of concrete vs. abstract words has recently been shown to be differentially affected by tDCS of the LH’s core language system: specifically, cathodal tDCS of Wernicke’s area was found to positively affect overnight consolidation of newly learnt abstract (but not concrete) words as opposed to sham (placebo) stimulation (Kurmakaeva et al., 2021). This supports the view of LH advantage for abstract semantics; however, as the RH Wernicke’s area homologue has not been stimulated in the same fashion, its role in abstract vs. concrete word acquisition and processing remains untapped. Moreover, the evidence of RH involvement in linguistic learning in general has so far been rather unclear. While some results indicated activity in right occipito-frontal networks linked to verbal learning in brain-damaged individuals (Blasi et al., 2002), there is no similar data for a healthy population. To the best of our knowledge, there are no studies attempting stimulation of RH homologues of core language cortices to explore their causal role in the acquisition of different semantic types. The present experiment was aimed as a step towards filling this gap. To assess the role of Wernicke’s right-hemispheric counterpart in word acquisition and processing, we used tDCS to stimulate this area before a contextual learning task, in which healthy participants were exposed to novel concrete and abstract words in the context of short stories in tightly controlled experimental settings.

tDCS is a non-invasive brain stimulation method that (similar to magnetic simulation) can be used to study causal relationships between brain areas and their functions. One advantage of tDCS is its putative ability to gradually modulate the resting membrane potentials of neurons, shifting the inhibition/excitation balance and therefore affecting plastic properties of neural networks (Blagovechtchenski et al., 2019). This, along with the method’s safety, non-invasiveness, ease of use, and low cost, has contributed to its growing popularity. Interestingly, cathodal and anodal polarities of such electric brain stimulation have been shown to differentially affect motor excitability, causing either inhibition (cathodal tDCS) or excitation (anodal tDCS) in the motor cortex, with some differential effects on the cognitive processing as well (Nitsche and Paulus, 2000, Flöel et al., 2008). However, tDCS of language areas has so far produced somewhat controversial results (see, e.g., Klaus & Schutter, 2018), which do not allow for clear differentiation of polarity effects. One possible explanation for this is the complex dynamic involvement of both excitatory and inhibitory neural processes in the brain’s cognitive operations. Therefore, even if cathodal stimulation may reduce cortical excitability, this does not necessarily disrupt cognitive functioning, as has been shown, for example, for working memory and attention (Demeter, 2016, Roux and Uhlhaas, 2014). To explore tDCS effects on right-hemispheric involvement in novel word acquisition, the present study used both anodal and cathodal stimulation of rW, as well as a sham control condition, in a between-group design to directly compare the effects of stimulation polarities.

Investigations of newly learnt, previously unfamiliar concrete and abstract words, rather than extant wordforms and meanings, have a crucial advantage of ruling out any confounds related to preexisting word properties (e.g., frequency or length, which can be matched precisely when creating new wordforms) and personal associations and experience of operating these words (Kurmakaeva et al., 2021). By using such an approach and matching physical and psycholinguistic stimulus parameters as well as the word acquisition regimes, one could objectively address both the basic behavioural (e.g., accuracy and speed of acquisition) and neural (brain areas involved) mechanisms that may underpin previously described differences between concrete and abstract semantics (such as the well-known concreteness advantage effect; Paivio, 1990, Schwanenflugel et al., 1992, Fliessbach et al., 2006) that still remain the subject of a debate. Based on the theories and empirical data mentioned above, we hypothesised that tDCS application over Wernicke's area RH homologue will have a differential impact on abstract vs. concrete word acquisition. This might become manifest as a relatively greater susceptibility of novel concrete word processing to the modulation of this area with tDCS, as opposed to the abstract words which have previously been posited to be predominantly supported by LH and should thus be less affected by rW tDCS.

On the other hand, there is indirect evidence that RH may also be involved in the processing of abstract semantics. When the entire LH is surgically removed in infancy, the development of the lexicon is supported by the RH, providing it with capability for recognising both abstract and concrete words alike (Ogden, 1996). Furthermore, the right hemisphere, particularly the RH’s Wernicke's area homologue, has been shown to play a crucial role in processing nonliteral language such as novel metaphors (Mashal et al., 2005), jokes and idioms (Brownell et al., 1983, Van Lancker and Kempler, 1987). Since figurative language involves a great degree of abstraction, one might argue that rW may be involved in acquisition of abstract semantics. Furthermore, as mentioned above, the right hemisphere is involved in social cognition, whose role in the comprehension of metaphors, jokes, and abstract concepts has been emphasised by different researchers. For instance, Fini and colleagues (2021) have shown that individuals required more social interaction when performing tasks using abstract concepts. In similar vein, a connection between right hemisphere and emotional processing may also promote its involvement with abstract language, as abstract words are known to have stronger affective associations than concrete ones at both behavioural (Vigliocco et al., 2014; Ponari, Norbury & Vigliocco, 2018) and neural levels (Etkin et al., 2006). Therefore, an alternative hypothesis is that tDCS of this area could influence the processing of both abstract and concrete words alike.

A further open question concerns long-lasting effects of tDCS on word acquisition, in particular, their sustainability after the stimulation session. An overnight sleep stage is believed to be necessary for consolidating newly established word memory traces (Davis and Gaskell, 2009, Dumay and Gaskell, 2007), transferring them from short-term to long-term memory (Rasch & Born, 2013). For instance, the role of sleep was demonstrated in the integration of newly learned words into pre-existing lexicon in both school children and adults (Dumay and Gaskell, 2007, Henderson et al., 2012). As mentioned above, cathodal Wernicke’s area tDCS positively affects overnight consolidation of new word memory circuits (both abstract and concrete ones; Kurmakaeva et al., 2021), which, in view of other evidence of RH’s involvement in language, suggests that tDCS over RH’s Wernicke's area homologue may also have similar effects. Such sustained effects would indicate the role of this area in word acquisition and consolidation; furthermore, if successful, this approach could potentially lead to improved speech rehabilitation techniques using tDCS, which have so far produced controversial results (Ulanov et al., 2019).

In sum, the experimental hypotheses of the present study are as follows:

In light of the available evidence of (a) RH’s, particularly rW’s, involvement in figurative language processing as well as social and emotional cognition, and (b) facilitatory effects of anodal tDCS on language learning, we hypothesised that anodal tDCS of the right-hemispheric homologue of Wernicke’s area may facilitate the acquisition of novel abstract words, which will be manifested in higher accuracy and faster reaction times in assessment tasks, as opposed to placebo stimulation. Alternatively, based on the available data of differential asymmetry of concrete vs. abstract word processing, we might expect that RH’s stimulation will predominantly enhance concrete rather than abstract words. Cathodal tDCS, in turn, might inhibit novel semantics acquisition.

As previous studies suggested long-term effects of Wernicke’s area tDCS on new word acquisition, we might expect similar findings for rW tDCS, which could be observed in larger differences between real stimulation and sham groups 24 h after the learning session than immediately after it.