Metaphors are an integral part of speech and enrich communication, but they also, as shown by Lakoff and Johnson ^[1], form our opinion about the world and the phenomena that surround us. For instance, we often use the metaphorical concept TIME IS MONEY (Lakoff and Johnson also introduced writing the concepts in capital letters, which is generally accepted in the literature dealing with (conceptual) metaphors). (e.g., “You’re wasting my time.”) or LOVE IS A JOURNEY (e.g., “This relationship is a dead-end street.”) (Examples taken from ^[1]). These concepts allow us to map knowledge about one domain, which is physical and more basic (money, travel), to a domain that is more complex and abstract (time, love).

Metaphors are used in speech and expressed with words, which makes them an area of interest in linguistic research, but they also represent a great challenge in the field of natural language processing. Precisely because they are an integral part of speech and of how we express ourselves, if we want to achieve further progress in natural language processing and artificial intelligence, metaphors are a phenomenon that needs to be studied in an interdisciplinary manner, so that we can understand their cognitive linguistic theory and recognize the reach of computational linguistics as well as machine and deep learning.

Since Lakoff and Johnson ^[1] published their Conceptual Metaphor Theory (CMT), in the book “Metaphors we live by” in 1980, this has been the dominant theory in metaphor research. The theory of the conceptual metaphor is considered to have been most influential in shifting the focus from words and expressions to the cognitive process, prompting the development of the cognitive theory of the metaphor, a subfield of cognitive linguistics that further developed new theories, criticism, and approaches ^[2].

While the theory of the conceptual metaphor deals with how the metaphor impacts our understanding of the world, linguistic metaphor research deals with the metaphor at the level of language, and how metaphors are expressed in language.

Below is an overview of the conceptual metaphor theory, followed by an overview of the linguistic metaphor.

In “Metaphors we live by”, Lakoff and Johnson present their theory stating that metaphors influence our thinking and actions, and that our conceptual system is metaphorical in nature. They support the theory with concepts such as ARGUMENT IS WAR (e.g., “Your claims are indefensible.”, “He attacked every weak point in my argument.”)–stating that it is “important to see that we don’t just talk about arguments in the terms of war”, but that we actually can “win or lose arguments”. Furthermore, the authors conclude that “the essence of metaphor is understanding and experiencing one kind of thing in terms of another”. In the specific example of ARGUMENT IS WAR, we comprehend ARGUMENT and talk about it by means of WAR. The authors support their theory by citing a series of concepts and examples from everyday life and communication (for example, TIME IS MONEY, LOVE IS A JOURNEY, MIND IS A MACHINE, THEORIES ARE BUILDINGS). At the core of the conceptual metaphor is the mapping of one domain (source) to another (target), as shown in the example ARGUMENT IS WAR, where WAR is the source and ARGUMENT the target domain. This mapping takes place not only at the linguistic level but also at the cognitive level, which affects our understanding of the world and the phenomena that surround us. Lakoff and Johnson focus precisely on the discovery of such concepts (in the mind) and primarily deal with conventional metaphors (cf. ^[3]), which are so common in everyday speech (and life) that we do not even notice them. It is important to distinguish among common, conventional metaphors and novel metaphors. It could be argued whether the conventional—novel metaphor distinction is a binary one or a scale, but what is interesting about novel metaphors is how some of them become conventional. Those that do are often accepted in many languages ^[3][4]—which means that these languages, apart from sharing the same metaphors on the linguistic level, share the same concepts, i.e., understanding of the world. Some of the concepts cited by Sulivan ^[4] that are applicable in several languages are, for example, describing a person’s intelligence through brightness, which we find, for example, in English “bright student” and Spanish “una persona brilliante”. Describing anger as a hot liquid (e.g., “let off steam”, “steaming at the ears”) is also a concept that we find in a number of languages—English, Japanese, Chinese and Hungarian. The existence of the same or similar concepts in different languages is further evidence for the conceptual metaphor theory. Another characteristic of the conceptual metaphor is its asymmetry, i.e., its unidirectionality—if we use the same example of ARGUMENT IS WAR, we can see that we can express argument using the experience of war, but we will not explain war through the experience of the argument. The same applies, for instance, to LOVE IS A JOURNEY—we do not talk about a journey using the experience of love. Lakoff and Johnson ^[1] explain this phenomenon by noting that concepts that are less clear and less concrete are explained through concepts that are more clear, concrete and grounded in our experience. Relying on the theory of the conceptual metaphor, Grady ^[5] in his doctoral thesis develops the theory of primary metaphor, which shows us how metaphors can be divided into primary and complex metaphors, which we create from primary metaphors. Primary metaphors represent simple patterns such as MORE IS UP (e.g., “My income rose last year”) (Example for MORE IS UP concept taken from ^[1]), and these are the atoms that make up molecules, i.e., complex metaphors ^[5][6]. Grady gives the example of the concept THEORIES ARE BUILDINGS and decomposes this concept into simpler and more primary concepts (atoms) making up a more complex concept (molecule)—ORGANIZATION IS PHYSICAL STRUCTURE and VIABILITY IS ERECTNESS. Lakoff and Johnson ^[7] accepted the primary metaphor theory and developed it further. The authors state that the theory of the integrated primary metaphor has four parts, i.e., four individual theories that they develop further:

• Theory of conflation ^[8] which proposes that children combine sensory and non-sensory experiences from a very early age, resulting in later understandings of metaphors such as “close friend”, “warm smile” and “big problem”.
• The theory of the primary metaphor ^[5][6], which states that all complex metaphors are “molecules” composed of “atomic” metaphorical parts, which we call primary metaphors.
• Neural theory of metaphor ^[9] that claims that the associations created during the conflation period are realized through permanent neural connections in the neural network that defines the conceptual domains.
• Theory of conceptual blending ^[10] which states that distant domains can be connected and create new deductions.

These four parts taken together form the integrated theory of the primary metaphor, which claims that people acquire a large system of primary metaphors automatically and unconsciously, in everyday and ordinary situations, from an early age, which leads to us thinking by using hundreds of primary metaphors ^[7]. The authors further present a short representative list containing 23 primary metaphors (AFFECTION IS WARMTH, IMPORTANT IS BIG, HAPPY IS UP, INTIMACY IS CLOSENESS, BAD IS STINKY, DIFFICULTIES ARE BURDENS, MORE IS UP, CATEGORIES ARE CONTAINERS, SIMILARITY IS CLOSENESS, LINEAR SCALES ARE PATHS, ORGANIZATION IS PHYSICAL STRUCTURE, HELP IS SUPPORT, TIME IS MOTION, STATES ARE LOCATIONS, CHANGE IS MOTION, ACTIONS ARE SELF-PROPELLED MOTIONS, CAUSES ARE PHYSICAL FORCES, RELATIONSHIPS ARE ENCLOSURES, CONTROL IS UP, KNOWING IS SEEING, UNDERSTANDING IS GRASPING, SEEING IS TOUCHING, PURPOSES ARE DESTINATIONS, PURPOSES ARE DESIRED OBJECTS). In an appendix to his doctoral dissertation ^[5], Grady also provides a list of primary metaphors, which he divides into 5 main categories: Atemporal relations; Quantity and degree; Time, action and event structure; Affect, evaluation and social relations; Thought and consciousness, which contain a total of about 100 primary metaphors, i.e., concepts. It should be emphasized that none of these lists is definitive. Grady ^[6] states that “humans everywhere share the basic patterns of experience that are reflected in primary metaphor” so it is possible to find these patterns in various languages, and primary metaphors are the same in different languages. For example, warmth is associated with affection which is considered as something good, and this concept has its roots in the warmth children feel when their parents hold them in their arms (e.g., “Warm welcome”, “Thank you for your warm words”). Lakoff and Johnson ^[7] state that primary metaphors are part of unconscious cognition, and that we acquire them unconsciously as part of the normal learning process that we have no influence over. This process is the same for all people, which points to the fact that the acquisition of primary metaphors is universal. The authors emphasize that they are not innate, but learned, and are manifested through language, i.e., words.

Lakoff and Johnson’s theory of the conceptual metaphor is undoubtedly extremely important and has greatly influenced metaphor research, but the authors do not deal with the linguistic metaphor, that is, with how the conceptual metaphor is expressed in language. Knowledge of the theory of the conceptual metaphor is certainly necessary in order to recognize or identify a metaphor in a sentence (or text), but a metaphor is expressed by language and in language (There are, for example, visual metaphors, but this is beyond the scope of this paper), and it is language that is the subject of research when trying to identify a metaphor.

When discussing the linguistic metaphor, in the context of creating a system for identifying metaphors, Shutova ^[11] believes that attention needs to be given to the degree of conventionality of the metaphor, the syntactic construction of metaphorical expressions and the level at which we observe or annotate the metaphor—at the level of a word or a sentence or at the level of the source and target domains within grammatical relations. ^[12]

By exploring metaphors in educational discourse, Cameron ^[12] provides an overview of the linguistic metaphor. On a data set of 26,613 words, Cameron finds 711 instances of the linguistic metaphor and concludes that the frequency of the metaphor is 27 metaphors per 1000 words. She also states that, by type of word, metaphors are most often expressed by verbs (47%, which includes verbs, phrasal verbs, and verb phrases), followed by prepositions (34%, which also includes prepositional phrases).

Furthermore, Shutova and Teufel ^[13] give a statistical overview of corpora that contains texts from six domains (literature, politics, news, sociology, speeches and scientific articles on literature). These authors state that metaphor expressed in verbs present 68% of all metaphors, which clearly indicates that metaphors are much more often expressed by verbs than by other types of words.