Introduction

Human senses of olfaction and gustation developed early, and the chemosensory system on which they rely conferred an evolutionary advantage by warning of dangers in the environment (Olofsson and Freiherr, 2019). In other words, the primary sensory experiences elicited by smell and taste contribute to people’s perception of their surroundings. Olfactory metaphors (such as ‘Something smells fishy.’) and gustatory metaphors (such as ‘Sweet memories.’) refer to linguistic phenomena where abstract meanings are conveyed through the senses of smell and taste. Various cross-cultural and within-cultural studies have explored the association between the abstract meanings and the basic meanings of olfactory and gustatory metaphors in different parts of speech (Basalaeva, 2019; Ibarretxe-Antuñano, 2000; Scott, 2023; Torres Soler, 2021; Zawisławska and Falkowska, 2018).

According to Conceptual Metaphor Theory (CMT), metaphors serve as both rhetorical devices in language and cognitive tools. They use concrete concepts (source domains, such as sense, motion, and space) to express and understand abstract and complex concepts (target domains, such as social status, morality) (Lakoff and Johnson, 1999, 2003). With the development of embodied cognition theory, metaphors are seen as reflecting the evolutionary development of human cognition and as a fundamental mode of thinking. Abstract concepts are formed through bodily perceptual experiences of the world (Barsalou, 1999; 2008). A database of Chinese olfactory and gustatory metaphors could be used to validate these theories by exploring how concrete senses are employed to express abstract concepts.

Researchers have widely studied the connection between the concrete sense of smell/taste and their mapped abstract metaphorical meanings (Lee and Schwarz, 2012; Hellmann et al., 2013; Lee et al., 2015; Ren et al., 2015; Wang et al., 2019). For instance, the concept of ‘sweet revenge’ links concrete sensory experiences to abstract social judgments (Hellmann et al., 2013). The gustatory experience of sweetness can influence the perception of a vengeful act as more lenient. This metaphor functions by linking a physical sensation (sweet taste) to an abstract idea (revenge). Conceptual metaphors involve general mappings, such as ‘Morality is cleanliness’ (Schnall et al., 2008), while idiomatic metaphors are specific phrases like ‘Revenge is sweet,’ requiring concurrent activation of sensory and abstract components (Hellmann et al., 2013). Despite these studies, there is a scarcity of behavioral and neuroimaging studies delving into the rich metaphorical nuances of smell and taste (Citron and Goldberg, 2014; Pomp et al., 2018; Zhang et al., 2023). As a result, the understanding of the mechanisms underlying olfactory and gustatory metaphors is limited.

Recent neuroimaging studies reveal that when comprehending sensory metaphors, the corresponding sensory brain areas are activated. For example, Lacey et al. (2012) examined the processing of the texture metaphor ‘She had a rough day’ and identified activation in the somatosensory areas responsible for texture representation and selection. In other words, the metaphorical expression of ‘rough day’ activated the brain regions associated with texture perception and simulation. Similarly, Citron and Goldberg (2014) and Pomp et al. (2018) conducted studies exploring the neural basis of gustatory and olfactory metaphors, respectively. The latter found activation in the secondary olfactory cortex but no activation in the primary olfactory cortex. This may be due to differences in stimulus materials used in the studies. For example, the stimuli used by Pomp et al. (2018) were not segmented by part of speech, influencing the activation patterns observed. According to the theory of embodied cognition, the activation of the primary olfactory cortex may be related to the accumulation of sniffing action experiences. This indicates that different types of lexical items (e.g., verbs vs. adjectives) might engage distinct neural mechanisms, showing the necessity for careful consideration in stimulus selection and categorization. Therefore, it is important to maintain strict control over stimulus materials when researching sensory metaphors to ensure accurate and comparable results.

Previous research has examined the impact of stimulus control on metaphorical language processing. Initial experiments suggested a specific role of the right hemisphere (RH) in metaphorical meaning processing (Anaki et al., 1998; Bottini et al., 1994). However, subsequent studies have not consistently replicated this finding (Lee and Dapretto, 2006; Mashal et al., 2009; Rapp et al., 2004), with Rapp et al. (2007) proposing that both hemispheres may participate in metaphorical meaning processing, and that factors beyond metaphoricity could also trigger right hemisphere (RH) activation. It’s worth noting that studies reporting RH involvement exhibit variability in terms of materials, methods, and implementation, potentially explaining the divergent results (Koller et al., 2021; Yang, 2014). This variability underscores the importance of considering psycholinguistic variables in the factor control of stimuli and the processing of metaphorical language (Koller et al., 2021), with related research exploring psycholinguistic variables such as familiarity, emotional valence, aptness and so on (Blasko and Connine, 1993; Cardillo et al., 2012; Citron and Goldberg, 2014; Damerall and Kellogg, 2016). Hence, this study has incorporated seven key variables, primarily aimed at standardizing metaphorical and literal descriptions of smell and taste. For more details regarding their Importance, impact on metaphor processing, and interpretation of the ratings, please refer to Table 1.

Table 1 Definitions of dimensions, presented to participants in the instructions for the test.
Full size table

Our study categorizes expressions into adjectives, nouns, and verbs, considering their notable metaphorical implications and the distinct cognitive abilities required for comprehension, which may involve diverse neural processing mechanisms (Cardillo et al., 2010). The most extensively studied type of metaphor, as discussed by Lakoff and Johnson, is the noun metaphor (Lakoff and Johnson, 2003). It can be comprehended through comparison, categorization processes, or a combination of both (Bowdle and Gentner, 2005; Glucksberg, 2003). Processing nominal metaphors, such as ‘Jealousy is a tumor,’ involves comparing the properties of the target term (‘jealousy’) with the base term (‘tumor’) for overlap. Categorization-based theories suggest that this processing entails constructing a category covering both the target and base terms, with the latter being more representative of the category’s meaning (Gokcesu, 2009). However, it is important to note that metaphors can also involve other parts of speech. For instance, in the expression ‘the stock soared,’ the verb carries the metaphor whose meaning is abstracted from its literal meaning (Chen et al., 2008). Verbal metaphors’ structure reflects both the general processing of actions and the specific processing of the verb itself.

Currently, no specific tool or database for olfactory and gustatory metaphors in Mandarin Chinese has been developed. Different cultures may attribute varying symbolic meanings to smells and tastes, influencing the use and interpretation of metaphors (Majid and Burenhult, 2014; Wnuk and Majid, 2014). Mandarin Chinese, with the largest number of native speakers worldwide, has a diverse range of olfactory and gustatory expressions rich in metaphorical meanings. Therefore, such a database would be useful. Moreover, previous studies based on Mandarin Chinese investigating gustatory metaphors primarily focus on materials associated with concrete object, such as ‘lemon’ and ‘sauerkraut’, rather than exploring the metaphorical aspects of ‘sour’ itself. In contrast, our database would include words related to qualities like ‘fragrant’ or ‘smell,’ rather than concrete objects like ‘cinnamon’ (Barrós-Loscertales et al., 2012; González et al., 2006).

The purpose of this study was to create a normative Chinese Olfactory and Gustatory Metaphor Database (COGMED) by gathering ratings from native Chinese speakers on metaphorical and literal expressions related to the olfactory and gustatory senses. The entries in the database were classified according to their part of speech (adjective, verb, or noun) and whether they were literal or metaphorical, in accordance with the established practices in previous event-related potential (ERP) or functional magnetic resonance imaging (fMRI) studies on metaphor processing (Coulson and Van Petten, 2002; Pomp et al., 2018; Pynte et al., 1996). All items were evaluated across seven psycholinguistic dimensions: familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness (please refer to Table 1). In the evaluation process, for example, when assessing the imaginability of the phrase 香艳的花朵 ‘fragrant flowers,’ subjects rated the ease with which the phrase evoked a visual image on a scale ranging from 1 (no image) to 5 (clear, direct image).

Overall, this study aimed to build a comprehensive, multisyntactic database of Chinese olfactory and gustatory metaphors, testing its psycholinguistic dimensions. We extracted, constructed, and normalized the target database from local texts. Then, we evaluated it on seven psycholinguistic dimensions, ultimately obtaining high reliability and correlations consistent with previous research. In the discussion section, we recommend ways of using the database.

Materials and methods

Participants

In total, 200 students were recruited from Southern Medical University in China (41 males, 159 females, age: 18.42 ± 0.86). The inclusion criteria required that all participants be native Mandarin Chinese speakers in good physical and mental health, with no history of brain trauma, nervous system disease, or olfactory/gustatory dysfunction. The evaluation survey questionnaire and methodology were examined, approved, and endorsed by the research ethics committee of Southern Medical University (No. 2022–17). All procedures performed in this study were in accordance with the ethical standards of the university. The documentary evidence is provided in Supplementary Information 3.

Materials

Extraction of source domain expressions

Metaphors are often viewed as a cognitive tool that transfers meaning from concrete concepts to abstract domains (Lakoff and Johnson, 1999; 2003). For instance, the metaphor ‘Jealousy is a tumor’ employs a familiar source domain (tumor) to enhance understanding of the less familiar target domain (jealousy). In our study, descriptions of smell and taste are employed as the source domains.

Ten single characters (five olfactory and five gustatory) were selected as they represent the fundamental concepts of smell and taste in Mandarin Chinese (Andersen et al., 2020; Cai, 1996; Chaudhari and Roper, 2010). The olfactory characters selected are 香 (xiang1) ‘fragrant,’ 臭 (chou4) ‘smelly,’ 腥 (xing1) ‘fishy,’ 膻 (shan1) ‘muttony,’ 臊 (sao1) ‘urine odor,’ and the gustatory characters are 酸 (suan1) ‘sour,’ 甜 (tian2) ‘sweet,’ 苦 (ku3) ‘bitter,’ 咸 (xian2) ‘salty,’ and 鲜 (xian1) ‘umami’. Furthermore, the character 辣 (la4) ‘spicy’ was included within the taste domain. Despite being associated with pain, as Just et al. (2007) demonstrated by using capsaicin to measure the trigeminal nerve threshold, we deemed it pertinent to explore this commonly used taste concept, as it is frequently discussed in everyday conversations concerning taste perception.

Collection of vocabulary expressions

Based on the identified olfactory and gustatory single characters, a search was conducted to confirm words related to the senses of smell and taste. Key sources included authoritative dictionaries and compendiums of Chinese characters, namely the Xinhua Dictionary, the Modern Chinese Dictionary, and the Shuowen Jiezi (a classical Chinese semantic dictionary). The Xinhua Dictionary, in particular, is widely accessible in contemporary China (Cai et al., 2022). The definitions in the Modern Chinese Dictionary are predominantly clear, albeit with some obscure exceptions (Li, 2011). The Shuowen Jiezi, as the inaugural Chinese dictionary, plays a pivotal role in elucidating the meanings of traditional Chinese words (Xu, 2004). Adjectives, verbs, and nouns were considered for inclusion in this process, and a total of 97 words (38 olfactory words and 59 gustatory words) were obtained. Table 2 shows examples of words and single characters, and Table 1 in the Supplementary Information 1 displays the frequencies of metaphorical vs. literal expressions and of adjectives, nouns, and verbs (obtained from the CCL corpus; Zhan et al., 2019).

Table 2 Examples of corpus material.
Full size table

Construction and normalization of stimuli

The CCL corpus (Center for Chinese Linguistics, 2019) is a national Chinese corpus in widespread use. The raw linguistic data are primarily derived from newspapers due to their relative authoritativeness (Zhan et al., 2019), and can be processed to meet the user’s intended purpose. As the CCL corpus covers older usages of Mandarin Chinese, it can be supplemented by web searches of more frequently used expressions to ensure that all data is up-to-date. Web searches employ text-based applications and Baidu for comprehensive coverage.

In the subsequent phase, utilizing the aforementioned olfactory and gustatory terms as keywords, we systematically gathered expressions including phrases, sentences, and idiomatic sayings containing these terms from the CCL corpus and the Internet. However, we encountered several inconsistencies in the CCL corpus, notably in syntax and context. Therefore, we retained the essential smell/taste words and their collocations to ensure that the individual expressions remained fundamentally consistent under various conditions (such as between literal and metaphorical or among adjective, noun, and verb expressions) to improve the match functions. For example, during our search in the CCL corpus, we modified the original phrase ‘The wind carries cranberries to the barbed wire while fragrant flowers adorn thistles’ to the simplified expression ‘fragrant flowers’.

Adjectival expressions comprised an olfactory/gustatory adjective and a corresponding noun (X + Y, two words). For example, in 香艳的花朵 ‘fragrant flowers,’ X denotes 香艳的 ‘fragrant,’ and Y denotes 花朵 ‘flowers’. Nominal expressions consisted of two nouns linked by a copula (X is Y, three words), as illustrated in 鼻子是器官 ‘The nose is an organ,’ where X represents鼻子 ‘the nose,’ and Y represents 器官 ‘the organ.’ In verbal expressions, olfactory/gustatory verbs integrated a subject and an object to form a coherent sentence (X v Y, three words). For instance, in 警犬嗅包裹 ‘Police dogs sniff parcels,’ v denotes 嗅 ‘sniff,’ X denotes警犬 ‘Police dogs,’ and Y denotes包裹 ‘parcels.’ (See Table 2 for more examples). Thus, the literal-metaphorical-abstract representations of these smell and taste concepts were all standardized as phrases or sentences consisting of two or three words and adhering to a uniform format across respective parts of speech (adjectives, verbs, and nouns).

Screening and specification

In this study, the initial collection pool of olfactory and gustatory items included over 400 entries each. To ensure the quality and comprehensibility of the items, ten psychology students were recruited to identify any entries they deemed incomprehensible or unfamiliar, leading to the removal of any sentences consistently categorized as such. Following this screening process, 319 olfactory and 352 gustatory items remained, each of which was divided into literal and metaphorical expressions. In the olfactory test, 245 items constituted adjectival phrases (145 were metaphorical [MI], 100 were literal [LI]), 52 items were verbal sentences (27 MI and 25 LI), and 22 were nominal sentences (15 MI and 7 LI). In the gustatory test, 270 items came from the adjectival phrases (178 MI and 92 LI), 30 were verbal sentences (17 MI and 13 LI), and 52 were nominal sentences (34 MI and 18 LI).

Procedure

This survey incorporated distinct olfactory and gustatory assessments, administered to the participants in an alternating odd-even sequence. The survey consisted of an informed consent page, basic personal information, textual instructions, and a scale from 1 (extremely bad) to 10 (extremely good) on which participants self-rated their senses of taste and smell. Participants also rated the remaining items on a 5-point Likert scale across seven dimensions: familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness (see Table 1).

The survey was conducted using a pen-and-paper format, without a specified time limit for completion. Participants took between 45 and 90 min to complete the questionnaire. In cases where responses were missing due to participant error or intentional omission, the researcher made efforts to contact the individuals and obtain the missing item scores.

Statistics

Statistical analysis was conducted using SPSS (version 20.0; IBM, Somers, NY). For brevity, only significant results are reported in the main text. A full report of the statistical results and the complete database for olfaction and gustation is available in Supplementary Information 2. The ratings for the seven variables were statistically analyzed. The Cronbach’s α coefficient was used to measure the reliability of the item ratings across participants. Correlations for dimensional scores were calculated based on the sense (olfactory and gustatory) and part of speech (noun/verb/adjective). Scores for variables such as gender, age, and self-perception were calculated for each dimension to explore any potential effects. To investigate the differences between metaphorical expressions and literal expressions, we employed either a paired t-test or the non-parametric Wilcoxon signed-rank test, depending on whether the differences exhibited normal distribution (indicated by a Shapiro-Wilk test with p < 0.05) or contained significant outliers.

Results

Data screening

A total of 200 individuals initially took part in the study, with one person dropping out at short notice, leaving 199 entrants. Data from the olfactory and gustatory questionnaires, completed by 97 and 102 participants respectively, were screened. To ensure compliance with instructions (Yao et al., 2017), we checked the data to establish that no entry contained 85% or more identical responses across all items. Next, means and standard deviations for each item were calculated. One participant from the olfactory group and two participants from the gustatory group, whose emotional valence ratings negatively correlated with item means, were excluded from the analysis. Additionally, we excluded one gustatory participant whose ratings showed poor correlation (r ≤ 0.10) with other participants assigned to the same item (Warriner et al., 2013; Bradley and Lang, 1999). Thus, four participants were removed, leaving a total of 195 responders, with 96 completing the olfactory test and 99 the gustatory test.

Rating for familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness

Table 3 shows the descriptive statistics for psycholinguistic dimensions: familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness. Figure 1 shows the distribution of scores for olfactory and gustatory questions on the seven dimensions. Table 2 in Supplementary Information 1 shows the items that correspond to the maximum and minimum values of each dimension throughout the entire database (for instance, the most/least familiar items), alongside the descriptive statistical values of these items.

Table 3 Descriptive statistics of the tests.
Full size table
Fig. 1: Box plots of the mean distribution of scores on the seven dimensions for the entries on olfaction and gustation.
figure 1

The information includes maximum, minimum, upper, and lower quartiles, median, and outliers.

Full size image

In addition to summary statistics for variables, we also consolidated the characteristics of all items in the other categories (olfactory–gustatory type, metaphorical–literal expression, and adjective–verb–noun expression) (see Table 4) and tested the differences between metaphorical and literal items across seven dimensions (See Table 5).

Table 4 Summary of stimulus characteristics by item type.
Full size table
Table 5 Comparisons between metaphorical stimulus and literal stimulus in different item types.
Full size table

Reliability

We calculated the inter-rater reliabilities (Cronbach’s alpha) of each item’s dimension ratings. For the olfactory database, the Cronbach’s alphas for the seven dimensions ratings demonstrated high reliability, ranging from 0.92 to 0.99 (values for familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness were 0.99, 0.99, 0.99, 0.92, 0.99, 0.99, and 0.98, respectively), yielding a mean of 0.98 (SD = 0.03). Similarly, for the gustatory database, the Cronbach’s alphas for the seven dimensions ratings ranged from 0.96 to 0.99 (values for familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness were 0.99, 0.99, 0.99, 0.96, 0.99, 0.99, and 0.98, respectively), with a mean of 0.99 (SD = 0.01).

Correlations between dimensions

Olfactory items

Spearman correlations were employed to assess the relations among dimensions for the ranked data means (see Fig. 2). A strong positive correlation was reported between the five dimensions of familiarity, meaningfulness, difficulty, imageability, and naturalness (for any two of these, r ≥ 0.86, ps < 0.001), indicating that more familiar olfactory items were rated as more meaningful, accessible, imaginative, and natural. Items rated as more meaningful were also perceived as more imaginative, accessible and natural; Items rated as easier it is to understand were perceived as more imaginative and natural; similarly, more imaginative items were associated with higher naturalness. Figurativeness exhibited moderate-to-strong negative correlations with familiarity (r = −0.57, p < 0.001), meaningfulness (r = −0.60, p < 0.001), difficulty (r = −0.71, p < 0.001), imageability (r = −0.66, p < 0.001) and naturalness (r = −0.62, p < 0.001), indicating that less figurative olfactory items were perceived as more familiar, meaningful, imaginative, and natural, and easier to comprehend.

Fig. 2: A heat map of the correlations between the olfactory score dimensions.
figure 2

Note. **p < 0.01, ***p < 0.001, ****p < 0.0001.

Full size image

Gustatory items

In general, as with the olfactory data, the gustatory dimensions were strongly and significantly correlated (see Fig. 3). The strongest positive correlations were observed between any two of the following dimensions: familiarity, meaningfulness, difficulty, imageability, and naturalness (0.76 ≤ r ≤ 0.93, ps < 0.001). Similar to the olfactory findings, negative correlations were noted between figurativeness and dimensions of meaningfulness (r = −0.19, p < 0.001), valence (r = −0.15, p = 0.006), difficulty (r = −0.28, p < 0.001), imageability (r = −0.23, p < 0.001), and naturalness (r = 0.11, p = 0.036).

Fig. 3: A Heat map of the correlations between the gustatory score dimensions.
figure 3

Note. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Full size image

Relations among dimensions within parts of speech

The seven-dimensional values collected for each item were correlated separately by sensory type (olfactory vs. gustatory) and part of speech (adjectival, verbal, and nominal; see Table 6) in the metaphorical expressions to further explore their relationships, which differed considerably. The most pronounced correlations between dimensions occurred in adjectival metaphors, while nominal and verbal forms exhibited weaker correlations, possibly attributable to the smaller number of items. Similarly, mirroring the patterns observed in the two sensory types, the dimensions of familiarity, meaningfulness, difficulty, imageability, and naturalness demonstrated overwhelmingly positive correlations across the three parts of speech.

Table 6 Correlation coefficients between item types.
Full size table

Effects of gender, age, and self-perception on dataset ratings

We calculated male and female participants’ mean ratings for the seven dimensions to investigate potential gender-related differences across the two sensory types and three parts of speech. There was no significant difference between the average ratings of male and female participants in the olfactory or the gustatory tests and age/self-perception were not correlated with the item ratings (ps > 0.05).

Discussion

This paper has reported on the development of COGMED, the first metaphorical database of independent gustatory and olfactory sensations in Mandarin Chinese. By assessing the psycholinguistic factors (familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness) known to influence language processing, COGMED provides researchers with reliable explanatory analyses of relevant factors. To better control for confounding factors and facilitate the matching of items, the database can be searched by part of speech, by the length of expressions, and by grammatical structure and metaphorical/literal keyword positions under the same part of speech.

We first measured the differences between the metaphorical and literal expressions on the seven psycholinguistic dimensions of familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness and explored the correlations between them. Our olfactory results (Table 4) indicate that metaphorical expressions received higher scores in figurativeness, while literal items were perceived as more positive, familiar, meaningful, imageable, natural, and easier to comprehend compared to metaphorical expressions. The gustatory results showed the same pattern as the olfactory results except for familiarity. Concerning the interrelations among these ratings, a significant positive correlation was observed between any two dimensions of familiarity, meaningfulness, difficulty, imageability, and naturalness in both the olfactory and gustatory data (Figs. 2 and 3). Previous studies have shown similar results: whereas the meanings of less familiar items may be more difficult to identify (Schmidt and Seger, 2009), previous research also shows that the qualities of imageability and naturalness are more complex aspects of all metaphorical items (Cardillo et al., 2010). Meanwhile, the negative correlation between figurativeness and other dimensions shows that the less metaphorical an expression is, the easier it is to understand. This is again consistent with previous studies, suggesting that metaphorical processes require more processing time and working memory due to the need for more semantic processing (Schmidt and Seger, 2009).

Comparing the metaphorical and literal differences across the three parts of speech (nouns, verbs, and adjectives) revealed occasional inconsistencies between the ratings and overall mean scores in the verbal and nominal expressions (Table 4). One possible explanation is that insufficient numbers of these types of expressions existed in the database. Compared to other dimensions, those of figurativeness and valence generally recorded the lowest ratings. This discrepancy may be attributed to the differing scoring rules for the two dimensions compared to others. Specifically, most ratings were based on a 5-point scale, with a score of 3 considered neutral. However, the figurativeness of an expression was only considered to be extremely low when it scored below 2; scores greater than or equal to 2 were regarded as increasingly metaphorical. The low mean validity scores may indicate the presence of a more balanced number of negative, neutral, and positive emotional expressions in the database.

The division of different parts of speech of items in COGMED is helpful to the study of sensory metaphor. In recent neuroimaging research, olfactory or gustatory stimuli showed activity in the primary and secondary gustatory cortices during the reading of gustatory metaphors (Citron and Goldberg, 2014), but this research has limitations, such as insignificant primary olfactory cortex activity due to unsegmented linguistic input by part of speech. Pomp et al. (2018) found that olfactory verbs and adjectives mainly activated the secondary olfactory cortex, with less activation in the primary cortex, possibly due to not comparing different lexical forms separately.

González et al. (2006) found that reading words with strong olfactory associations (compared to control words) prompted bilateral activation in primary olfactory areas (for example, the piriform cortex) and the right amygdala, suggesting that comprehending these words somatopically activates the motor and premotor cortices. This is consistent with the theory that abstract concepts are understood by analogy with sensations and actions (Lakoff and Johnson, 1999, 2003). Conversely, Aziz-Zadeh et al. (2006) found somatotopic activation in the premotor cortex for literal action sentences, but not for metaphorical phrases such as ‘biting off more than you can chew’. Our comparative use of adjectival, verbal, and nominal expressions for smell and taste from this database aided a clearer investigation of the relationship between sensory and abstract concepts.

In the present study, ratings for both olfactory and gustatory items exhibited enhanced inter-rater reliability, with reliability coefficients for all dimensions exceeding 0.92 and average reliability factors surpassing 0.98. Compared to the other six dimensions, valence ratings demonstrated lower inter-rater reliability for both olfactory and gustatory items, probably due to different parts of speech. However, valence reliability remained between 0.92 and 0.98, indicating a fairly high level of reliability.

When applying COGMED, metaphorical–literal types and smell–taste models serve as a starting point. After determining the defining values of the dimensions required by the experiment, differences caused by different metaphor types or modal items were compared. A second approach is to start with dimensions. For example, items with different familiarity levels can be used to explore the influence of novelty on metaphorical comprehension. The last and most direct application of COGMED is to select or adapt materials according to the purpose of experiments related to olfactory or gustatory metaphors. For instance, items in COGMED provide minimal expressions, allowing subsequent researchers to add contextual background as needed. This control over the difficulty, richness, and contextuality of the reading stimulus in experiments can be tailored to the specific objectives of the study.

In ERP experiments, all sentences consist of two to three words, with adjectives, verbs, and nouns expressed in them with the same focal word positions, facilitating comparison and manipulation of the data (Canal et al., 2022). However, the number of metaphor types in COGMED is not balanced, with far fewer olfactory and gustatory verbs and nouns in the database. Nevertheless, these considerations do not reduce COGMED’s applicability to a range of potential uses.

Our database is composed of the same number of characters in most items, is grammatically consistent, and keywords are controlled in the same positions in the same type of speech, even in the adjectives and noun items, which are different types of speech with sensory-related words controlled in the opening two positions. Therefore, they may be suitable for highly time-sensitive methods, such as eye movements, pupil dilation, and electroencephalogram/magnetoencephalogram (EEG/MEG) use. They are also suitable for other methods, such as fMRI (functional magnetic resonance imaging) and TMS (transcranial magnetic stimulation). The database project, with its simple grammar and small number of words, is also suitable for different target groups, such as people of different ages and those with psychiatric disorders.

Conclusion

COGMED provides ratings for familiarity, meaningfulness, figurativeness, valence, difficulty, imageability, and naturalness for 671 olfactory and gustatory noun, verb, and adjective items. We hope that COGMED can be used for subsequent research, such as corpus-based analysis, and as material for studies on cognitive mechanisms, thereby promoting the study of olfactory and gustation metaphors.