Positive emotion of self-referential contexts could facilitate adult’s novel word learning: An fNIRS study

Highlights

• Emotions inherent in self or other referential contexts pose important impacts on adults’ novel word learning.

• Besides the finding of self-promoting effects, there were a prominent self-positivity bias during novel word learning.

• The DLPFC and inferior frontal gyrus played key roles in novel word learning under contexts.

Abstract

Learning words through contextual inference is a key way to enlarge one’s vocabulary especially for adults. However, few studies focused on the effects of different information contained in contexts on novel word learning. The present study used behavioral and fNIRS techniques to investigate the influences of positive, neutral and negative emotions inherent in self-related or other-related referential contexts. Participants were asked to perform a semantic consistency and a source judgment task after learning the relations between novel words and concepts in different contexts. The results showed that self-reference during lexical encoding could promote word learning generally. More importantly, there existed a self-positivity bias which is manifested in the significant interactions between contextual emotions and referential value. These interactions are related to the neural activities of the DLPFC and IFG. These results revealed the contextual information’s integrative contributions to semantic meaning acquisition and episodic source memory related with novel word learning.

Introduction

Words are basic elements of language, so learning words is the first step to acquire a new language (Lesage, Nailer, & Miall, 2016), which becomes increasingly necessary under the current circumstances of globalization. Therefore, more and more researchers start to focus on investigating the mechanisms of word learning, especially the learning process under different contexts (Chen et al., 2014, Dautriche and Chemla, 2014, Liu, Ding, Li, & Yang, 2019, Mestres-Missé, Muente, & Rodriguez-Fornells, 2009, Wang & Mintz, 2018), since contextual learning largely mimics the natural language learning situations. Besides predictive semantic information, contexts can also contain emotional connotations or have self-referencing value for language learners. Hence the learning mechanisms of novel words are probably influenced by the contextual information. Therefore, the current study aimed at revealing the effects and neuro-mechanisms of contextual emotions and self-reference on novel word learning by employing behavioral measurement and functional near-infrared spectroscopy (fNIRS) technique, which is a method to access the cortical activities by measuring the concentration change of oxygenated hemoglobin (HbO), deoxygenated hemoglobin (HHb) and their sum total in the cerebral cortices.

Previous studies have shown that readers could use contextual information to infer a novel word’s meaning (Chen et al., 2014, Dautriche and Chemla, 2014, Mestres-Missé, Camara, Rodriguez-Fornells, Rotte, & Muente, 2008, Mestres-Missé, Muente, & Rodriguez-Fornells, 2009). Through monitoring readers’ eye movements during reading pairs of sentences containing a target word, a correspondent context and a target-related word, Chaffin and his colleagues (2001) found that participants would spend more time on the informative contexts every time the target words were novel, and if the contexts were neutral or uninformative, they would turn to the target-related words so as to infer the meanings of the novel target words. Similarly, Mestres-Missé and her colleagues (2008) asked participants to infer the meanings of novel words embedded in contexts with different restrictions, and found that participants could only infer the meaning of new words across semantically consistent contexts but not across semantically inconsistent contexts. Furthermore, researchers compared the learning effects of concrete words versus abstract words and nouns versus verbs using similar learning paradigm, and found that learning different types of words implicated different brain areas (Mestres-Missé, Muente, & Rodriguez-Fornells, 2009, Mestres-Missé, Rodriguez-Fornells, & Muente, 2010). Thus, it’s repeatedly found that adults can learn different conceptual meaning from contexts.

As for the effects of emotions contained in contexts, most of research focused on existing word processing other than novel words learning (Guo, Li, Zhang, Cui, & Wei, 2018, Smith et al., 2004, Zhang, Liu, An, Yang, & Wang, 2015). Emotional information was mainly manipulated into two dimensions: valence and arousal. Arousal is a continuous value varied from boredom/calm to anxiety/excitement, while valence is a continuous value varied from unpleasant to pleasant (Bradley, Greenwald, Petry, & Lang, 1992). The two dimensions of emotions can play important roles in learning (Kensinger, 2009). In a ERPs study, participants were asked to study neutral words superimposed onto emotional images during encoding session and performed a classic old/new judgment task in testing session. The ERPs data collected during encoding session revealed that items presented in the positive and negative high-arousing contexts relative to neutral context elicited more positive ERP deflections. During retrieval phase, FN400 old/new effects occurred for words learned in the negative low-arousing, positive, and neutral contexts but not in the negative high-arousing context, while LPC old/new effects occurred for all conditions of contexts. The results showed that emotional contexts could pose impacts on both familiarity and recollection processes, and emotional arousal irrespective of valence seemed to impair the memorization of words generally (Zhang et al., 2015). A recent study has also shown that participants paid less attention to words presented in the high-arousing contexts compared with the neutral contexts during word identification, suggesting that it might be effortful to overcome the interference of arousal contexts during the process of memory formation (Guo et al., 2018). However, Maratos and Rugg (2001) observed a larger old/new ERP effect for words studied in negative compared to neutral sentences when contextual information could be retrieved incidentally. Therefore, it seemed that both arousal and valence of emotional contexts have significant effects on cognitive processing for words.

Using fMRI techniques, Maratos, Dolan, Morris, Henson, and Rugg (2001) continued to investigate the dynamic activities of neural networks correlated with neutral word learning under emotional contexts. Participants were asked to perform an old/new recognition task after learning emotionally neutral words presented in sentential contexts that were either negative, neutral or positive. The results showed that the recognition of words presented in negative relative to neutral contexts was associated with increased activities in right dorsolateral prefrontal cortex, left amygdala and hippocampus, right lingual gyrus and posterior cingulate cortex. The dorsolateral prefrontal cortex is a neural base of general executive function in language processing (Raboyeau, Marcotte, Adrover-Roig, & Ansaldo, 2010), and is a part of attentional control network (Botvinick, 2007). The activations of amygdala, orbitofrontal cortex and hippocampus may reflect the retrieval and recognition of emotional contexts and items (Dolan et al., 2000, Maratos, Dolan, Morris, Henson, & Rugg, 2001), while the activations of anterior temporal cortex and posterior cingulate cortex may be related to emotional experience and episodic memory (Mesulam, 1985, Maddock, 1999). Similarly, Smith and his colleagues (2004) found that retrieving neutral images encoded in emotional contexts elicited enhanced activations of brain regions involved in episodic memory. Collectively, brain areas correlated with emotion processing, cognitive control, semantic and episodic memory probably are neural substrates for word learning under emotional contexts.

In stark contrast to abundant researches into the influential mechanisms of contextual emotions on existing words processing, there are few studies concerning about novel word learning. Although some studies confirmed that novel words’ concept could be acquired through different sentential contexts (such as Chen et al., 2014, Mestres-Missé, Muente, & Rodriguez-Fornells, 2014, Mestres-Missé, Rodriguez-Fornells, & Muente, 2010, Qiao & Forster, 2013; Ye, Mestres-Missé, Rodriguez-Fornells, & Muente, 2011), they did not take the emotions of contexts into consideration. Using behavior and fMRI techniques, recent studies have examined the influences of individual’s emotional states and sentential emotional information on novel word learning. The results showed that negative contexts impeded the learning process of novel words, and contextual emotions moderated the brain functional changes induced by novel words learning (Guo et al., 2018, Guo et al., 2018). Furthermore, Snefjella, Lana, and Kuperman (2020) investigated whether novel words could acquire emotional connotations based on linguistic contexts and whether these connotations will affect the quality of word learning. During learning phases, they asked participants to learn nine novel words presented in positive, neutral or negative contexts, and tested their learning quality of novel words’ orthographic and semantic information. The results showed that participants could learn both the emotional connotations, forms and semantic information of novel words from the contexts, and participants learned novel words more accurately in the positive contexts compared to the neutral and negative contexts. Therefore, it seems that positive contexts probably facilitate the word learning but negative contexts impede.

Regarding the impeding effects of negative contexts on words learning, some researchers thought that individuals were automatically alert to negative contexts, since the information contained in negative contexts is more urgent and may threaten a person’s survival. Consequently, negative contexts would take up a lot of attention resources, so that individual’s performances in processing words in negative contexts are relatively poor (Pratto & John, 1991). Studies have also shown that individuals are more sensitive to negative information (Cacioppo & Berntson, 1994) and react to it more strongly (Taylor, 1991), which may cause individuals to spend more attention on negative contexts, thereby ignoring words. With regard to the facilitative roles of positive contexts in learning, the broaden-and-build theory can give us a reasonable explanation (Fredrickson, 2001, Fredrickson, 2013). It’s proposed that positive emotions could broaden an individual's momentary thought-action repertoires (e.g., creativity, flexibility, high efficiency), and then build the individual's personal resources (intellectual, physical, psychological and social resources), thereby bring long-term benefits to the individual (Fredrickson, 2001, Fredrickson, 2013). Taken together, negative and positive emotion probably pose opposite effects on word learning through narrowing or broadening persons’ attention resources.

Self-referencing value is also a key property of learning contexts. Self-reference effects refer to a better memory performance on self-relevant words and the function of self seems to be a superordinate schema in encoding process (Rogers, Kuiper, & Kirker, 1977). Klein (2012) argued that self-reference could strengthen semantic links among materials, so individuals could automatically organize self-relevant materials. Many studies consistently documented self-reference effects, showing that self-relevant stimuli were more elaborate and processed faster (eg., Fan et al., 2013, Tacikowski et al., 2011). Researchers also studied self-reference effects in learning activities and found that self-reference played a key role in memory formation, organization and retrieval (Conway & Dewhurst, 2010). D'Ailly and his colleagues (1997) required children to solve mathematical problems in different referential conditions and found that in self-referential condition, children were less likely to repeat the questions and their performances were better than that in other-referential condition, indicating that self-referential encoding helped children find solutions. These findings were echoed by the Cunningham and his colleagues’ study (2018), in which children tended to learn self-owned shapes better than shapes owned by experimenter or un-owned. As for novel word learning, children showed 20% higher spelling accuracy for a novel word when it was compiled in a sentence with themselves as the subject of the sentence than with another meaningless words as the subject. It’s proposed that the self-relevant information is stored by children in their self-schema, which is connected with the learning words, stimulates learners’ interests and motivation, and makes the learning goals meaningful, thereby helps them to maintain and retrieve words (Kihlstrom, Beer, & Klein, 2003).

The medial prefrontal cortex tends to be implicated in self-referential processing (Tacikowski et al., 2011). Imafuku and his colleagues (2014) employed fNIRS to detect the brain activities of infants when they heard names and found that their own names induced greater activations in the medial prefrontal cortex. Northoff and his colleagues (2006) summarized 27 studies on self-reference and found that self-referential processing was closely related to the midline cortex, which was consistent with another study (Murray, Marie, & Martin, 2012). Additionally, studies suggested that right inferior prefrontal gyrus was crucial to self-evaluation (Morita et al., 2008) and self-related stimuli (Sugiura et al., 2000). Some researchers indicated that the inferior prefrontal gyrus was a part of the frontal-parietal attentional control network and played a crucial role in cognitive inhibition (Aron, Robbins, & Poldrack, 2014).

Self-reference and contextual emotion might jointly affect word processing. Fossati and his colleagues (2003) asked participants to judge whether a negative or positive personality trait could describe themselves and found that the right dorsomedial prefrontal cortex was uniquely activated in self-referential condition regardless of the valence of the words. They argued that the activations were not probably related to emotional contexts per se but were attributed to the self-referential processing of the emotional stimuli. In addition, researchers found that the positive adjectives encoded in reference to the self were better recalled than negative ones only in a retrieval task entailing cognitive control (D'Argembeau, Comblain, & Van der Linden, 2005). Moreover, researchers proposed a self-positivity bias which referred to that most individuals prefer to link self-relevant information to positive emotion (Pahl & Eiser, 2005). A recent fNIRS study detected brain activities when participants were reporting corresponding emotional feelings on evaluating art works from the perspective of themselves or artists. Oxyhemoglobin (HbO₂) analysis showed a significant activation in the rostrate prefrontal cortex in other-reference condition while Deoxyhemoglobin (HHb) results displayed an activation of this region in self-positive condition (Kreplin & Fairclough, 2015). Above mentioned studies underscored the role of the prefrontal cortex in processing referential and emotional information, so the present study will focus on observing activities in this region.

Based on above reviews, the emotional information and self-referential value in contexts are very important factors for language processing and learning. However, to our knowledge, there is no study to examine the potential interactive influences of these two factors on novel words learning. Novel word learning necessitates mapping a new concept to a specific linguistic symbol, and self-referential might well facilitate this process since self-reference can strengthen semantic links among materials (Klein, 2012). Given the bias of self-reference to positive information, we expected that there would be significant interactions between contextual reference and emotions.

Hence, reference and emotions of contexts were orthogonally manipulated to explore their interactive effects on novel word learning with behavioral and fNIRS techniques in the current study. Participants were required to learn the meanings of novel words which could be inferred from sentential contexts during learning session. During test session, they were asked to perform a semantic consistency task and a source judgement task. The semantic consistency task required participants to determine whether the learned novel words were consistent with the corresponding real words so as to check the learning quality of semantic information. We expected that there would be a self-referential advantage on learning novel words compared with other-referential condition especially for words learned in positive contexts. Furthermore, self-reference might involve enhanced activations in the prefrontal cortex compared to other-reference, and the emotional effects might be more prominent in self-referential contexts. Episodic memory is an ability to remember the bindings of the to-be-recalled items and the aspects of contexts (Andrews et al., 2020, Ngo, Horner, Newcombe, & Olson, 2019). Since self-referential contexts probably activate self-schema of participants, it’s reasonable to expect that self-reference would facilitate the memory of learning background. Therefore, after the semantic consistency judgment task, participants were also asked to perform a source judgment task, in which they were asked to recall the referential contexts in which the novel words were learned. We predicted that individuals could better recall the reference information of novel words in the self-referential and positive contexts.