The goal of this study was to explore the nature of antonymy as a semantic relation of opposition from the perspective of Cognitive Linguistics as an inter and multidisciplinary theoretical framework. The study wanted to show that antonymy is actually a relation of opposite meaning on the conceptual level, and that it functions as a prototype-based category. The reasoning behind this aim stems from the long tradition of the notion of opposition in Western scientific thought. INTRODUCTION The earliest preserved written records show that ancient Greek philosophers included relations of opposition in their treatises on the most various aspects of human preoccupation. Philosophers that came before Aristotle frequently described reality in terms of opposite notions. Pythagoras and his followers believed that reality was based on ten principles, which were formed as lexical pairs with opposite meaning, such as left-right, male-female, light-dark, good-evil, etc. Heraclitus also believed that opposition formed the basis of reality, but framed his claim in terms of the notion that “everything flows”, which became his Doctrine of Flux and Unity of Opposites. With this Doctrine Heraclitus proposed that opposites exist in everything at the same time, but that only one member of the pair can rule at one point in time. So, a man has the potential for being both healthy and sick, but can only be one at one moment in time. Parmenides also used relations of opposition to describe reality as he saw it. He also saw reality as being governed by opposites, such as day-night, light-dark, and reality-illusion. Plato used opposition to describe his view of reality, which included opposite notions being in a type of relation that allows them to transform into one-another, which is quite similar to the view of Heraclitus. According to Plato, everything that is small can become big, everything that is just can become unjust, etc. There is a zero value between each pair of opposites that allows them to undergo this transformation, which greatly resembles the traditional view of antonymy as being a gradable phenomenon. In his writings collected under the titles Metaphysics, Categories and On Interpretation, Aristotle looks back on the thoughts of his predecessors and includes opposition as one of the fundamental building blocks of our understanding of reality. His use of opposition for describing various aspects of reality served to form logic as a scientific tool, which is embodied in the square of opposition that is still used to distinguish among the various kinds of opposite meaning. With the square of opposition, Aristotle was able to differentiate between what we refer to in linguistics as binary and non-binary, and gradable and ungradable phenomena. The thoughts of Ancient Greek philosophers, and of Aristotle especially, were perpetuated through Western thought in the writings of various Christian philosophers, such as Thomas of Aquinas, and later on in the writings of famous philosophers of the Enlightenment, such as Kant and Schellig. Ancient, medieval, and modern philosophy proved to be of great significance for our current understanding of antonymy. The thoughts of generations of philosophers influenced the most significant thinkers in linguistics, especially Ogden (1932), who wrote the first book in English about opposition. This book presents an important stepping stone in the development of our understanding of antonymy, as it influenced the writing of Lyons (1977) and Cruse (1986) on the English language, as well as many other linguists writing about other European languages (e.g. Šarić wrote about antonymy in Croatian in 2007). Ogden (1932), Lyons (1977), and Cruse (1986) are credited with the first classifications of opposition in English, with Ogden’s classification being more general, while those proposed by Lyons and Cruse are linguistic in nature. The most commonly referenced classification distinguishes binary from non-binary opposition (e.g. big-small, days of the week), gradable from ungradable (hot-cold, man-woman), and morphologically related from unrelated (friendlyunfriendly, friendly-hostile) pairs of opposites. This type of classification is based on logical relations between members of the pairs, representing the traditional view of semantic relations of opposition taken up by Structural Semantics. This view focuses on lexical items and their interrelationships as they exist in the vocabulary of a language. However, this view, albeit providing a structured classification of the various phenomena that encompass opposite meaning, does not include real language use, remaining solely in the realm of theoretical vocabulary. Language use is of crucial importance for understanding the connection between language and thought, and this connection reflects the one between intralinguistic and extralinguistic phenomena. It is in this segment that the theoretical framework of Cognitive Linguistics comes in, because it stresses the importance of real language use and the native speaker. THEORETICAL FRAMEWORK For our understanding of antonymy, we included the general tenets of Cognitive Linguistics which strive to explain that language and thought are interrelated, as well as influenced by convention (Lakoff 1987, Langacker 1987, Fillmore 1988, Žic Fuchs 1991). Accordingly, we framed our understanding of this phenomenon through the notions of conceptualization, which operates on the basis of the connection between language and extralinguistic knowledge, categorization, which enables the structuring of knowledge via similarities and differences, prototypes and schemas, which are culturally based, convention, which dictates the phenomena that will be conceptualized, and non-modularism, which says that language is a cognitive capacity that works alongside other cognitive capacities. All of these notions together form a view of language that is inseparable from thought and experience, which includes concrete and abstract concepts, all submerged in the culture of the speakers (Wierzbicka 1992). Previous explorations of antonymy from the perspective of this framework (Jones 2002, Murphy 2003, Paradis 2011, Jones et al. 2012) have been valuable in their insights on the way opposite meaning is structured in language use, as well as in highlighting that it primarily resides on the conceptual level. These insights, however, mostly focused on adjectives, which were traditionally seen as true antonyms. Opposite meaning, though, can be expressed in all parts of speech. Furthermore, it has been shown that opposite meaning is expressed through specific syntactic structures (Jones 2002, Mettinger 2004). This proved to be crucial in our exploration of antonymy on a general scale, as we proposed that it is expressed via constructions (Goldberg 1995, 2006), which we refer to as antonymyc. Another question that was left unanswered by previous work has to do with the cultural specificities of antonymy. We believe that these could only be studied by comparing distinct linguistic systems that are products of different cultural contexts and cultural cognition (Sharifian 2011). We can see that the traditional and cognitive view of antonymy have many similarities regarding the basic principles of opposition. Both start from the notion that antonymy is a dichotomous relation between two lexemes of opposite meanings, which has a linguistic reality. The traditional view focuses on the vocabulary, while the cognitive view focuses on the conceptual level and its reflection in language use. However, the questions that remain unanswered, and to which we have tried to provide answers in this study, pertain to the scope of antonymy, its basic structure, its basic meanings, and its basic use in real language production. The scope refers to the fact that opposition permeates all parts of speech. All opposition in language shares a basic conceptual structure, which enables the expression of basic opposite meanings, which is why we decided to name the entirety of the phenomenon antonymy. Furthermore, antonymy in this sense is expressed through very specific constructions, which are part of its conceptual structure. The question of cultural specificity has not been posed so far, which prompted us to examine it through the comparison of English and Croatian. Since these two languages are typologically and culturally different, we believed that some aspects of antonymy would reflect some of their cultural differences. What we expected to find was that antonymy in general is a primarily conceptual relation which is reflected in language. We believed that it has a basic categorical structure with more and less prototypical members, which can be used to express several general opposite meanings via specific constructions. We also assumed that the more prototypical members of the category would be similar in English and Croatian, while the less prototypical members would be more culturally entrenched and, therefore, different in the two languages. METHODS Considering the questions and assumptions posed in the study, we devised a three-part methodology which would enable us to explore the underlined queries: a corpus analysis followed up by questionnaires, and finally an fMRI study of neuro-biological activity in the human brain while processing antonymy. Unfortunately, the final part of the study could not be conducted due to the COVID-19 pandemic, but will be conducted as a separate study when proper safety conditions are met. RESULTS The first part of the study included a thorough corpus analysis of both English and Croatian. The corpora used for the study were COCA for English and HrWaC for Croatian. The study began with a general search of all pairs of lexemes with opposite meanings that were found in the literature used for this study. Of these pairs, we selected those that proved to yield objective results, including pairs with higher and lower frequencies in both corpora. Once the main pairs were selected, we ran another search of the pairs, as well as of the individual members of the pairs. The goal was to see first what kinds of syntactic enviroments the pairs themselves were used in, and second whether the same kinds of environments were used with one member of the pair and a different other member. This part of the study yielded interesting results as it exposed some cultural differences between the English and Croatian categories of antonyms. The corpus analysis also revealed that both more and less frequent antonyms appeared in specific antonymyc constructions. The number of constructions rose with the frequency of the pairs. The frequency of pairs and accompanying antonymic constructions further revealed that there seem to be four basic opposite meanings that can be expressed with pairs of antonyms. Also, the consistent use of antonymic constructions with antonyms pointed to a specific basic structure of antonymy on the conceptual level. We found that pairs of antonyms, regardless of their part of speech, were structured as conceptual/semantic wholes that comprise the knowledge about the entirety of the subject. For example, the pair male-female forms a conceptual/semantic whole that denotes all the knowledge and experience, both linguistic and non-linguistic, that a speaker has about a human being, and which stems from the general convention of dividing humans into men and women. The two lexemes, male and female, form the two opposite poles of the conceptual/semantic whole, and represent a meeting point between knowledge of the language and knowledge of the world. This knowledge can be expressed through specific antonymic constructions (such as ‘male and female’, ‘male or female’, ‘neither male nor female’), which are stored in the speaker’s conceptual system as abstractions (‘X and Y’, ‘X or Y’, ‘neither X nor Y’), and which lend themselves to any other pair of lexemes with opposite meanings depending on their frequency. The same occurs in Croatian (‘muško ili žensko’ = ‘X ili Y’). There are four types of conceptual/semantic wholes: spatial-temporal relations (such as up-down or gore-dolje), qualitative designation (such as hot-cold or toplo-hladno), action (such as come-go or doćiotići), and entities and phenomena (such as man-woman or muškarac-žena). Each conceptual/semantic whole seems to prefer a set of antonymic constructions. The conceptual/semantic whole of qualitative designation has the widest set of antonymic constructions at its disposal, while actions have the narrowest set. The conceptual structure of antonymy allows speakers to express four basic opposite meanings: 1) both poles of the conceptual/semantic whole are expressed, 2) one pole of the conceptual/semantic whole is expressed, but not the other, 3) both poles of the conceptual/semantic whole are expressed, but one more than the other, and 4) neither of the poles of the conceptual/semantic whole is expressed. Each of the four basic meanings prefers specific antonymic constructions. For example, the first meaning prefers the antonymic construction ‘X and Y’, the second meaning prefers ‘X or Y’, the third meaning ‘more X than Y’, and the fourth meaning ‘neither X nor Y’. The same pattern was observed in Croatian. What we observed based on these findings is that antonymy is structured on the conceptual level as a prototype-based category whose members are included through conventionalized knowledge of the individual pairings and antonymic constructions. More prototypical members are more frequent in the corpora and can be found in a wider set of constructions, which can be used to express more basic opposite meanings. Less prototypical members are highly culturally determined and differ between the two languages. They are found in fewer antonymic constructions and are limited in the number of basic opposite meanings they can express. For example, both English and Croatian antonymy has the pairs man-woman/muškarac-žena as highly prototypical. As a less prototypical member of the category, English has the pair hotcool, while Croatian has the pair muškarac-pudlica. Furthermore, we also found some cultural differences between the two languages in terms of the secondary or metaphorical meanings of antonyms. For example, the pair black-white is mostly used in the English corpus in its metaphorical meaning denoting the color of one’s skin. In Croatian, the majority of meanings pertains to the primary sense of crno-bijelo, that denoting color terms. Furthermore, we observed that gradability as an aspect of antonymy can be found even in those pairs that have traditionally been defined as non-antonymy (e.g. complementaries and other types of contrast). For example, traditionally defined complementaries such as man-woman/muškarac-žena, boygirl/dječak-djevojčica and day-night/noć-dan all occur in gradable antonymic constructions (‘from X to Y’, ‘turns X into Y’). Therefore, we see gradability as an aspect of opposite meaning that can be used in any conceptual/semantic whole if the speaker has the need to express it. The second part of the study included questionnaires which were filled out online by native speakers of English and Croatian. The questionnaires were comprised of four tasks, each devised to explore an aspect of antonymy uncovered in the corpus analysis. The first task asked examinees to determine the second member of the proposed pair of antonyms. Examples included more and less prototypical pairs. The aim of the task was to see which pairs speakers saw as prototypical. The second task asked examinees to choose one among three pairs of lexemes which represented the best example of opposite meaning. Again, pairs included both more and less prototypical combinations. The aim of this task was to see which pairs of lexemes speakers would choose as the best example of antonymy when given a choice. The third task asked examinees to use provided pairs of lexemes in sentences in such a way that would clearly show their opposite meanings. All examples included more prototypical pairs of antonyms. The aim of this task was to see if examinees would use antonymic constructions in their sentences to express oppositeness of meaning. The fourth task asked examinees to provide the lexeme that they believed would be the proper opposite pair to the underlined word in the given sentence. All sentences were borrowed from corpora in order to preserve natural language use patterns. Examples of antonyms included more and less prototypical pairs. The aim of this task was to see if the presence of context and antonymic constructions would enable examinees to recognize pairs of antonyms regardless of their level of prototypicality. Our findings correspond to the results of the corpus analysis, except for two observed phenomena. The first and second tasks showed that examinees mostly agree on the pairs of antonyms they feel to be prototypical. The third task showed that some examinees used antonymic constructions to express opposite meaning, while others chose a descriptive method to do so, which was unexpected. The fourth task showed that while examinees had no difficulty in providing a member of a prototypical pair of antonyms with added context and an antonymic construction, when faced with an unconventionalized proposition, neither the context nor the antonymic construction proved to be sufficient in determining the pair of antonyms. This proves that antonymic constructions cannot function on their own to express opposite meaning. The speaker needs to have both knowledge of the language and knowledge of the world in order to use and recognize antonymy, which is reflected in its basic conceptual structure. CONCLUSION Based on the evidence provided by the corpus analysis and questionnaires, we are able to conclude that antonymy does function as a conceptual prototype-based category whose members are added according to conventionalized knowledge of the language and knowledge of the world. This conventionalized knowledge is influenced by the cultural context of the speakers, which means that some aspects of antonymy can differ from language to language. This coincides with the basic principles of the cognitive linguistic theoretical framework. The structure of the category of antonymy includes conceptual/semantic wholes and antonymic constructions which the speaker relies on to express four basic opposite meanings. The scope of antonymy encompasses all parts of speech which are evident in the types of conceptual/semantic wholes, each of which prefers a set of antonymic constructions aimed at expressing the four basic opposite meanings. Also, gradability should be seen as an aspect of the category of antonymy in general, as it can lend itself to various types of conceptual/semantic wholes, and not just those traditionally defined as antonymy proper. What we have shown in this study is an evolution of thought about opposite meaning from Ancient Greece to contemporary times, which points to the importance of this phenomenon for human thought and language. We have also shown that, when it comes to the subject of antonymy, Structural Semantics and Cognitive Linguistics follow this same evolutionary trajectory. Methods such as corpus research and questionnaires are useful for bottom-up research that begins with language use. However, in order to further investigate the subject, a multidisciplinary approach would open new methodological vistas and enable us to answer questions of an even more basic nature, such as what happens in the human brain when processing antonymy. Considering the cognitive linguistic preoccupation with language and thought, this would be an interesting step forward.