Research on dynamic categorization of word meaning: review and prospect

2.1 Perspectives on word meaning research

Word meaning encompasses both the dictionary-defined sense of a term and the contextual meaning generated within specific usage scenarios. These traditional associations are precisely what the dictionary captures (Bill 2023). This field of study, known as lexical semantics, falls under the broader umbrella of semantics. Lexical semantics is an academic discipline that studies the meanings of words. In linguistics, it significantly overlaps with what is traditionally known as lexicology, which is the comprehensive study of a language’s vocabulary, including its morphology, etymology, and topics related to social, regional, and dialectal variations (Cruse et al. 2002; Geeraerts 2010; Hanks 2007). Although word meaning has been discussed in Western literature since at least the time of Aristotle, lexical semantics as a field of study developed in the 19th century. At that time, lexical semantics had a historical-philological orientation, focusing mainly on etymology and the classification of how word meanings change over time. Early lexical semantics viewed word meanings as mental entities; they were considered thoughts, and changes in meaning over time were seen as the result of psychological processes (Geeraerts 2010). In the 20th century, new ideas were introduced by proponents of the structuralist movement, particularly Ferdinand de Saussure (1959). Contemporary semantics is characterized by diverse and influential paradigms, including structural semantics, interpretive semantics, generative semantics, and cognitive semantics. Each paradigm offers unique insights into how meaning is structured and understood, reflecting the evolution of semantic theory over time. Structural semantics predominantly adopts a synchronic approach, which, at a given time, emphasizes the composition and aggregation of words within a language system. However, it often gives limited consideration to extralinguistic factors such as social factors, physical factors, and speech contexts, which can influence the word meaning. According to structuralists, language is an autonomous internal system of relationships between words, organized based on lexical fields (Cruse 1986; Lehrer 1974). Interpretive semantics incorporates component analysis into the study of linguistic meaning, breaking down words into their constituent semantic features. At the end of the 20th century, a perspective emerged that opposed structuralism. Structuralism was replaced by generative and cognitive approaches. Katz and Fodor (1963) achieved a breakthrough with generative semantics. The main aim of generative lexical semantics research was to develop a logical formalism, either to deconstruct word meanings as per Katz and Fodor (Jackendoff 1983, 1990) or to construct word meanings akin to generative lexicon theory (Pustejovsky 2012). However, the methods of segmentation employed are often criticized for being arbitrary and subjective, leading to potential inconsistencies in analysis. Generative semantics, while providing a formalistic and rule-based framework for understanding meaning, tends to overlook the contextual and social dimensions of word meaning. This approach focuses on the syntactic generation of meaning, often neglecting how meaning is influenced by real-world usage and social interaction. In response to these limitations, George Lakoff, a prominent figure in generative semantics, identified numerous unresolved issues, prompting him to develop cognitive semantics as an alternative approach. Cognitive semantics seeks to address the shortcomings of previous paradigms by incorporating insights from psychology and cognitive science, emphasizing how meaning is constructed and understood in the human mind.

Over the past four decades, cognitive linguistics has emerged as a leading theoretical linguistic school globally, emphasizing the centrality of semantics in language understanding. This approach has revolutionized semantic research by highlighting the importance of mental processes and conceptual structures in the formation and interpretation of meaning. Cognitive linguistics evolved from the inheritance and critique of Transformational-Generative Grammar school, marking a significant paradigm shift that propelled semantic research into a new era of cognitive exploration. This reassessment questioned the adequacy of formalistic approaches and introduced a more holistic view of language as intertwined with cognitive processes. Furthermore, this paradigm shift has led to a gradual progression in the investigation of language, transferring from distant and external perspectives to closer, more internal views that consider the cognitive mechanisms underpinning language use. Researchers now focus on how linguistic meaning is rooted in everyday experiences and embodied cognition. However, the proliferation of semantic schools, coupled with the intricate and interdisciplinary nature of semantic research, has led to a lack of systematic coherence in the study of word meaning. This fragmentation often results in disparate and unconnected research findings, hindering the development of a unified theoretical framework. Xu (1997), as noted by Cao (2004: Preface II), observed that “Semantic works generally lack systematization, with chapters often lacking intrinsic connections”. This observation highlights the need for a more integrated approach to semantic research, one that can bridge the gaps between different schools of thought and provide a comprehensive understanding of word meaning.

2.2 Classification of lexical semantics

In linguistics, the emergence of derived word meanings plays a significant role in the formation of semantic types. However, linguistics lacks a universally accepted classification of word meanings. The classification of lexical semantics has long been a crucial issue. Scholars have attempted to classify word meanings, proposing different methods. Mirtojiev (2010) introduced a classified method from multiple perspectives, including semantic development, modes of semantic expression, word combinations, and types of meanings in lexical formation. He categorized word meanings into two types: root meanings and artificial meanings. Root meanings are the inherited meanings of words, including both the meanings derived from root words and those found in artificial words. Artificial words are initially created by humans and cannot be divided into morphemes from the perspective of modern languages. Mirtojiev argued that the inherited meanings of artificial words should also be considered root meanings. Furthermore, he emphasized the importance of affixes in the semantic structure of words, indicating that an artificial word’s semantic structure may include one or more meanings derived from the root meaning. In summary, lexical semantics can be classified from different perspectives, including semantic development, modes of expression, word combinations, and types in lexical formation. The primary importance of lexical derivation in determining semantic types, the expression of referents, the combinatory possibilities of words in semantic expression, and the typification of meanings in word formation should be central to this focus.

Takhirjonovna (2023: 1065) classified word meanings from the following perspectives: (1) Based on the development of word meanings. (2) Based on the expression of the essential meanings of lexicons. (3) Based on the word combinations to which the word meanings belong. (4) Based on the typification of word meanings in word formation. The classification of word meanings aids in the study of semantic derivation and is crucial for understanding the nature and characteristics of word meanings, reflecting the dynamic development process of semantics.

2.3 Cognitive semantics and word meaning

In traditional grammar, language is viewed as an external object, a set of prescriptive rules and norms to be followed. In contrast, a structuralist linguistics treats language as an observable phenomenon, focusing on the underlying structures and systems that govern linguistic behavior. Transformational-Generative Grammar, pioneered by Noam Chomsky, considers language to be a reflection of innate mental constructs and universal grammar. Cognitive linguistics, on the other hand, treats language as a manifestation of the cognitive system, emphasizing the role of experience, context, and embodied cognition in shaping linguistic knowledge. Within the framework of cognitive linguistics, lexicons and expressions directly map onto conceptual structures. Language is an integral part of human cognition. The function of the lexicon is to evoke conceptual patterns within the cognitive system. There is no algorithmic linguistic intermediary between language expressions and meanings. The cognitive approach to meaning is usage-based (Cruse 2002; Langacker 1999; Tomasello 2003). Scholars such as Shu (2002, 2008, 2013), Wen (2013, 2014), and Wang (2014) have provided comprehensive and systematic descriptions of semantics from a cognitive perspective. Their works have significantly advanced our understanding of how cognitive processes underpin the construction and interpretation of meaning in language. Building on these scholars’contributions, we summarize the fundamental theories of cognitive semantics in word meaning research. This framework encompasses “one adherence”, “two commitments”, “three views” and “four dimensions”, each representing a core aspect of cognitive semantics theory. “One adherence” refers to cognitive linguistics’ commitment to the “usage-based” approach to language acquisition. This approach posits that individuals do not possess innate language acquisition mechanisms, as proposed by Chomsky’s generative semantics. Instead, language skills are developed and refined through usage and experience, relying on general cognitive abilities. This view emphasizes the importance of context, social interaction, and pragmatic use in language learning.

In research on word meaning, cognitive semantics posits that word meanings are fundamentally grounded in empirical perception. As Katzir and Ariel (2024) stressed, in the cognitive-functionalist tradition, meanings are represented as rich conceptual structures, often including detailed information about how language users experience the world. This perspective emphasizes that our understanding of words and their meanings is fundamentally linked to human physiological structures and sensory experiences. This grounding in physical perception forms the basis for how we interpret and categorize lexical items in our cognitive framework. Ancient philosophies embraced the notion of “taken from body nearly and from materials far”, indicating that our ancestors formulated explanations and descriptions based on their immediate sensory experiences and perceptual capabilities. This principle reflects the deep connection between physical experience and linguistic meaning, highlighting how early humans used their bodily interactions with the environment to develop and refine their linguistic expressions. Secondly, individuals acquire word meanings of words through social interaction and communication, a process that is abstract and context-dependent. As Li (2024) noted, semantic changes are likely to be the byproduct of word meaning contextualization. Language learners and users employ various linguistic strategies to comprehend and articulate meanings, gradually deriving patterns from individual instances. These patterns are then reinforced and consolidated within their cognitive frameworks through repeated social and linguistic interactions, facilitating a shared understanding of lexical items within a community. Through repeated linguistic interactions, these patterns consolidate within individuals’ cognitive frameworks, fostering a shared understanding of word meanings across contexts. Furthermore, from the perspective of ecolinguistics, language and ecological environments interact with each other. By examining vocabulary through the concept of lexical niche, a theoretical foundation is provided for the sustainable development of vocabulary (Mo and Xiao 2024). Consequently, our understanding of word meanings evolves alongside changes in the ecological environment. This process ensures that word meaning is both dynamic and adaptable, reflecting changes in usage and context over time. Thirdly, the semantic attributes of certain words undergo transformations – such as stabilization, enrichment, or depletion – depending on their frequency and usage context. High-frequency words may become semantically enriched as they grasp additional nuances and meanings, while low-frequency words might experience semantic depletion, losing some of their original connotations. This dynamic aspect of word meaning highlights the fluid nature of language and its responsiveness to social and communicative needs.

The field of cognitive linguistics is guided by two fundamental commitments. The first commitment emphasizes the identification of general principles that can theoretically describe linguistic phenomena, particularly in the realm of word meaning research. This commitment seeks to uncover universal patterns and rules that govern how word meanings are constructed, interpreted, and communicated across different languages and cultures. The second commitment asserts that every linguistic principle must align with and reflect the existing understanding of human cognition as established by other branches of cognitive science, including philosophy, psychology, artificial intelligence, and neuroscience. This interdisciplinary approach ensures that linguistic theories are grounded in a comprehensive and scientifically validated understanding of the human mind and its cognitive processes (Pleyer and Hartmann 2024).

The “three views” in cognitive linguistics – experiential view, attentional view, and prominence view – offer distinct perspectives for studying language. In An Introduction to Cognitive Linguistics, Chen and Wen (2003) highlighted these views as central to the cognitive linguistic approach. The experiential view holds that language research should be empirical, grounded in actual usage and human experience, rather than relying solely on formal logic, objective definitions, and theoretical construction. This view posits that understanding language requires an examination of how it is used in real-life contexts, reflecting the cognitive and perceptual experiences of its users.

In research on word meaning, the experiential view posits three main points. Firstly, it suggests that the construction of word meaning attributes can be analyzed through the utilization of categorization theory. Word meaning cannot be solely defined by necessary and sufficient conditions or entirely explained by a set of semantic features. Instead, words or concepts are stored in the brain as typical categories, with their semantic attributes having various degrees of categorical membership maintained by family resemblance. Secondly, as Zhou and Huang (2024) presented, metaphors, as a cognitive tool, provide a robust framework for interpreting the numerous new things, concepts, information, and contexts that arise in our daily lives. The theory of conceptual metaphor marks a cognitive turn in metaphor research. CMT mainly deals with conceptual domains. A conceptual domain has been defined as a coherent area of conceptualization against which semantic units may be characterized, and it is relevant when it comes to the dissection of a conceptual metaphor. A conceptual metaphor takes place when our cognitive mechanisms are used to express and understand a conceptual domain in terms of another (Salvador 2022). Conceptual metaphors are based on mappings between similar categories. With systematicity and coherence, it therefore can be used to explain the emergence of new meanings in words. Conceptual metaphors emphasize deep cognitive modes of thinking, in which the essence of the metaphor is found not in the direct representation of linguistic forms, but in the intricate conceptual structures that underlie these forms (Lin 2024). Thirdly, the experiential view posits that conceptual metonymy is based on mappings of proximity. Metonymy allows for the understanding of unfamiliar, inanimate, and non-interactive word meaning through the use of familiar, animate, and interactive word meaning (Wen 2014). In other words, “alternative relations in languages are based on adjacency” (Xiao and Wen 2023: 159). Image schemas in cognitive semantics, such as container schemas and part-whole schemas, embrace experiential, gestalt, and dynamic characteristics. These schemas significantly impact word meaning extension, as variations in image schemas can lead to meaning extension, and metaphorical projections of image schemas can also cause a meaning extension. Once activated by related words, the default values in schemas are triggered, creating relevant connections with other schemas. In translation studies, Y. Zhou and H. Zhou (2022) suggested that the generation and construction process of “spiritual essence” involves activating schemas through language, mapping to certain mental activities, and constructing real or mental images based on what is read (Shu 2008: 29), followed by a series of cognitive processes involving schema matching, selection, processing, inference, and meaning acquisition.

The attentional view posits that the aspects of events upon which we focus are reflected in language. In the realm of word meaning research, a significant contribution is frame semantics. Frame theory was proposed by Fillmore and Atkins (1992), and Ungerer and Schmid (2001) regarded frames as cognitive models that represent knowledge and beliefs about specific situations. When determining word meaning, it is essential to consider not only the semantic features listed in dictionaries but also cultural background, encyclopedic knowledge, and life experiences. The meaning of words should be analyzed within a specific frame, where words and frames interact; words can activate frames, and frames are necessary for determining word meaning. Concepts closely related to frames include schema, script, and idealized cognitive model (ICM). These four concepts assist in predicting word meaning based on our knowledge structures and experiential backgrounds.

The prominence view asserts that the selection and arrangement of information within linguistic structures is determined by the degree of prominence. This view emphasizes that certain elements within a sentence or phrase are made more prominent or salient based on their relevance or importance to the overall meaning being conveyed. In the context of word meaning research, the prominence view manifests in two primary ways. This dual manifestation helps to explain how different lexical elements are highlighted or foregrounded to convey specific meanings or relationships. Firstly, there is the figure-ground theory, which is crucial for explaining the meaning of spatial prepositions. This theory posits that in any given scene, there is a “figure”, which represents the smaller, moving object with a clearer structure, and a “ground”, which represents the less differentiated, blurry background. The figure is the focal point of attention, often the smaller, moving, and more clearly defined object, while the ground provides the contextual background, often less distinct and more static. Langacker and his colleagues referred to the figure as the “trajector” and the ground as the “landmark”. This terminology underscores the dynamic relationship between the two elements, where the trajector is perceived in relation to the landmark. They utilized this theory to elaborate on the semantics of prepositions such as “in”, “out”, “over”, “on”, “under”, “up”, and “down”, providing a detailed account of how these prepositions convey spatial relationships. The figure-ground theory helps to describe the relationships among the different senses of these prepositions by highlighting how the position and movement of the trajector relates to the landmark. Secondly, there is the base-highlight theory or ground-highlight theory, which plays a crucial role in word meaning comprehension. This theory posits that certain aspects of a word or phrase are highlighted or made more prominent based on their relevance to the context. This highlighting effect is one of the four dimensions of comprehension, along with figure-ground theory, context, and experiential background.

The four dimensions refer to the levels of detail, perspective, base-active zone, and mental scanning that speakers rely on in semantic or word meaning comprehension. Cognitive semantics emphasizes human subjectivity in research on word meaning. The four dimensions of cognitive comprehension are not objective descriptions of internal features of word but powerful tools for understanding word meaning in different contexts. Cognitive linguists believe there is no direct mapping between words and scenes. A scene can be “comprehended” through these four dimensions and encoded in various ways to achieve different conceptualizations. The existence of multiple comprehension dimensions explains why word meaning dynamically develops over time and across contexts.

Cognitive semantics provides a systematic approach to research on word meaning. Categorization theory provides a theoretical foundation for constructing semantic attributes and explains semantic polysemy and vagueness. Metaphor, metonymy, and image schemas offer mechanisms for extending word meaning. Frame theory enables the prediction of dynamic word meaning development, while the figure-ground theory and base-highlight or ground-highlight theory provide frameworks for understanding word meaning in different contexts. These classic cognitive semantic theories contribute to a comprehensive understanding of word meaning research.

2.4 Word meaning and categorization theory

In cognitive semantics, categorization theory is considered the most fundamental theory for word meaning research. The study of categorization mechanisms is primarily represented by two theories: classical categorization theory and prototype categorization theory. Aristotle, the representative of classical categorization theory, put forward the following basic assumptions: Categories are defined by the conjunction of necessary and sufficient features, and an entity can only be considered a member of the category if it meets all these features; The features of entities are binary; they either belong to a category or they do not, with no intermediate states; there are clear boundaries between categories; all members of a category hold equal status. Rosch, the representative of prototype theory, argued that: Prototype members are the cognitive reference points of a category, and members within a category share family resemblances; category members include central and peripheral members, and their status is unequal; the boundaries of categories are not clear but fuzzy. Classical categorization theory plays a dominant role in 20th-century linguistic research, particularly in phonology, syntax, and semantics, which were also based on binary distinctions. Componential analysis (CA) in semantic research also adopted the binary opposition principle through a feature analysis to construct structuralist semantics. However, when applied to explain more social phenomena (including language), classical categorization theory struggled due to its lack of explanatory power, as most philosophical and social science categories do not exhibit binarity.

Classical categorization theory has three main shortcomings: Some categories are difficult to define by necessary and sufficient conditions; there are status differences among members; this theory cannot explain the fuzziness of category boundaries. Prototype theory addressed these issues, significantly advancing the study of categorization. In a sense, prototype categorization theory is the foundation of cognitive linguistics.

In recent years, the Chinese linguistic community has made significant strides by introducing categorization theory, particularly prototype categorization theory, into the study of word meaning. This integration has yielded notable results, contributing to a deeper understanding of how words are categorized and understood in various contexts. For instance, Yang and Zhang (1998) provided an introduction and comprehensive review of categorization theory. Zhao and Zhou (2000) utilized prototype theory to analyze the semantic categories of polysemous words, offering insights into how multiple meanings of a single word are related. Yu and Huang (2008) employed prototype theory to examine semantic fuzziness, shedding light on the gradations of meaning that occur within language. Additionally, Kuang (2014) explored the application of categorization theory in English vocabulary learning and teaching. His research provided practical strategies for educators to enhance vocabulary acquisition through the understanding of categorization principles. By combining the findings from Wen’s (2014) seminal work, the Cognitive Foundation of Language, we can summarize the categorization theory-related research on word meaning categories. This comprehensive overview provides a framework for understanding the advancements and methodologies used in this area of study.

First, the prototype theory provides a cognitive foundation for describing semantic categories and fuzziness. The prototype theory reveals the internal structural characteristics of semantic categories, indicating they are referenced to prototypes and connected through family resemblances. The prototype is the basis and core of a semantic category, while the periphery is composed of non-typical members. The “typical–non-typical” schema structure of categories proposed by Zhou (2000) and the “semantic category prototypicality” proposed by Wu and Ji (2004) explained the prototypical characteristics of semantic categories, showing that the prototype is at the center, members have different degrees of membership, and category boundaries are fuzzy. Furthermore, prototype theory can be used to analyze fuzzy hedges.

Second, the family resemblance of semantic categories provides a scientific and systematic basis for explaining polysemy. Both the diachronic evolution of semantics and the semantic categories formed by the synchronic meanings of polysemous words contain one or several core meanings (prototype meanings), from which other meanings are extended and developed, exhibiting family resemblances among the senses. Zhang (2009) believed that prototype categorization theory is particularly significant for the study of polysemy. The meanings of polysemous words can form typical structures, family resemblance structures, or lexical networks. For example, the meanings of “change” include: “alter”; “modify, vary”; “transfer (between vehicles)”; “change one’s clothes”; “substitute”; “exchange”; “voice change (especially for boys during puberty)”; “loose change”. According to family resemblance theory, the first meaning is central and the most typical meaning, while the seventh and eighth meanings are at the periphery of the semantic category. Thus, “alter” is the most representative meaning within the semantic category of “change,” with other meanings connected to it through family resemblance, forming a meaning chain or network in a radial or chain-like pattern. Peripheral meanings are the most abstract and gradually deviate from the prototype.

Third, categorization level theory provides a basis for explaining word meaning relations and the hierarchy of semantic fields. Rosch et al. (1976) divided categories into superordinate, basic, and subordinate levels. They believed that these three levels differ in informativeness and usefulness and played different roles in human cognition. For example, among the three words “vehicle”, “bus” and “city bus”, “vehicle” represented the superordinate level, “bus” basic, and “city bus” subordinate, respectively. Lakoff (1987) and Wang (2014) argued that the basic level of categorization exhibits perceptual completeness, cognitive recognition, priority in status registration, consistency in behavioral response, commonality in linguistic communication, effectiveness of relevant cues, and organization of knowledge and thinking. Superordinate categories emphasize and aggregate functional attributes. Only certain functional or purposive attributes of basic level categories are deemed essential or prominent and are extracted or borrowed by superordinate categories. For instance, “bus” is categorized as a “vehicle” to emphasize its function of transporting people or goods. Subordinate level categories possess unique attributes that are typically not shared with other categories, such as “city bus”.

Over the past twenty years, research on categorization in semantics has shown that most studies on prototype theory, family resemblance theory, and categorization level theory tend to focus on the static and synchronic aspects of lexical expansion. However, these studies often overlook the diachronic and instantaneous aspects of the dynamic nature of word meaning categories. Language development is a dynamic process. From an individual perspective, words reflect a centripetal movement towards personalized meanings, while language as a whole represents the centrifugal divergence of different users’ experiences toward a more objective concept. This dynamic also carries specific cultural characteristics, encompassing broader perceptual features and the practices of insiders within that culture (Zhou 2023). Therefore, this process also means that categories are constantly changing. Therefore, further in-depth research on the categorization theory in research on word meaning is necessary.

4.1 Deficiencies in dynamic categorization

The theory of dynamic categorization plays a pivotal role in the study of word meaning change. It provides a framework for understanding how meanings evolve over time and across different contexts, reflecting the fluid and adaptive nature of language. However, an overview of research on dynamic categorization reveals that scholars have yet to systematically construct a robust theoretical framework for this concept. Their understanding of the “dynamic nature” of categories remains fragmented and lacks comprehensiveness, which is mainly reflected in the following aspects:

First, the definition and research scope of dynamic categorization needs to be clarified. In terms of the definition of dynamic categorization, previous studies have rarely provided a unified and coherent description of “dynamic categorization”. Moreover, there is still a certain limitation in the acknowledgment of the dynamic nature of categories. Hopper, Thompson, Lakoff and Sinclair, for example, primarily focused on dynamic categorization at the level of word meaning.

Second, the theoretical system of dynamic categorization needs to be improved. In terms of the depth of theoretical research, the following issues need to be further explored and improved: First, the manifestation of the phenomenon of dynamic categorization of word meaning in different linguistic dimensions needs to be improved. The dynamic categorization of semantic categories involves the dynamic development of meanings at the morpheme, sentence, and discourse levels (Zeng and Wen 2018). At the lexical level, dynamic categorization is mainly reflected in the dynamic categorization of semantic features and semantic relations. At the sentence level, dynamic categorization is primarily manifested in the dynamic categorization of grammatical items and structural relations. At the discourse level, dynamic categorization is mainly reflected in the dynamic categorization of evaluative meanings and modal meanings of words. Second, the study of the working mechanisms of dynamic categorization of word meanings in different linguistic dimensions needs to be further detailed. Many scholars, both domestic and international, such as Liu Zhengguang, have explored the mechanisms of decategorization and recategorization. However, the following issues remain unresolved: (a) Dynamic categorization not only includes decategorization and recategorization, which is merely the third stage of the dynamic categorization of word meanings. It also includes the gradual changes within categories and interactions and crossings between categories in the first and second stages, and the mechanisms of these two stages need to be studied. (b) Scholars have already proposed that metaphor and metonymy are mechanisms of decategorization and recategorization. Barcelona et al. (2018) discussed the trend of metonymy research, arguing that metonymy research is a complex task, and some common questions in metonymy research are still unanswered, such as how to distinguish between metonymy and metaphor. In fact, besides metaphor and metonymy, we find that the mechanisms of dynamic categorization of word meanings also include base/profile. Additionally, metaphors can be further classified into conceptual metaphors, category metaphors, and grammatical metaphors, while metonymies can be classified into conceptual metonymies and grammatical metonymies. How these three major mechanisms play different roles at the levels of word meaning, sentence meaning, and discourse meaning will be one of the focuses of dynamic categorization theory research. Our future research should strive to improve the theory of dynamic categorization and systematize it.

4.2 Dynamic categorization perspective in research on word meaning

From the perspective of dynamic categorization, research on word meaning should prioritize the investigation of the types and processes involved in the dynamic categorization of word meaning. Building upon the work of previous scholars, we propose the following avenues for exploration:

First, it is imperative to clarify the motivations behind the phenomenon of dynamic categorization of word meaning. Bai and Wang (2013) observed that “the generation of conceptual recategorization is both the result of the dynamic adjustment of the lexical system and a concrete manifestation of the cognitive operations of language users.” This suggests that the dynamic development of categories is influenced by both internal and external factors, making the specific motivations for dynamic categorization worthy of further study from both subjective and objective perspectives.

Second, conducting systematic empirical research on the types and processes of dynamic categorization of word meaning based on corpus data analysis is crucial. Semantic categories can be classified into individual categories and composite categories. Based on the discussion by Ungerer and Schmid (2006), individual categories refer to basic, independent cognitive units that are typically associated with our direct experiences, such as “chair”, “tree”, or “dog”. These categories often have a central prototype and fuzzy boundaries among their members, but they all share some common attributes or features, which allows us to classify them together. Composite categories refer to the complex concepts formed by cognitively combining two or more different individual categories. This combination is not only based on literal meaning but also involves metaphor, metonymy, or other forms of conceptual integration. For example, a “wheelchair” is not just a combination of the core components “wheel” and “chair”, but it is also associated with other concepts such as “hospital” and “patient”, thus constructing a richer and more specific concept. Consequently, the investigation of dynamic categorization of word meaning should commence with these two types: the dynamic categorization of word meaning within individual categories and composite categories. Prior research on dynamic categorization of word meaning has not adequately distinguished the disparities between these two structures in relation to dynamic categorization. Furthermore, while both individual and composite categories may undergo gradual changes in the categories, interactions between categories, and decategorization, they are featured distinct inherent characteristics. Initially, it is imperative to broaden the research scope of dynamic categorization of word meaning within individual categories. Scholars’ research on the dynamic categorization at the level of word meaning mainly focuses on gradual change in the category of word meaning. Geeraerts, Ungerer and Schmid view the dynamic nature of categories as changes in the “prototype”, or “prototype shift” and “prototype splits”. This provides a new perspective for the study of semantic categories, but there are also shortcomings: the internal structure of categories involves more than just prototypes. Categories are composed of prototypes, category family members, and category boundaries. Changes in prototypes, the increase or decrease of family members, or adjustments to category boundaries all play an indispensable role in the change of word meaning; categories are divided into three levels: superordinate level, basic level, and subordinate level. Transitions among these levels also affect the formation or understanding of word meaning. Zeng (2020, 2021) has conducted pertinent research on the dynamic categorization of word meaning within individual categories through the analysis of examples. She has proposed that the study of gradual change in the category of word meaning should encompass dynamic changes in category boundaries, the dynamic construction of category prototypes, and dynamic adjustments of category levels. Furthermore, the dynamic development of word meaning entails not only the adjustment of internal structures within categories but also interactions between similar or adjacent categories, and even decategorization. Therefore, our study of semantic categories should not be confined to gradual changes in word meaning categories but should also consider the influence of adjacent and similar categories on a given semantic category, as well as more radical changes such as decategorization.

Third, the research on the dynamic categorization of word meaning in composite categories needs to be strengthened. Zhang (2007) discussed compound word formation and conceptual category compounds in his book Meaning and Cognitive Categorization. He distinguished between two concepts: category composition and category confusion. He provided an explanation of the dynamic development of composite categories and analyzed the attribution of category compounds. However, Zhang did not address the following issues: Firstly, what are the methods of category composition, and what are the characteristics of word meaning in composite categories after composition? Secondly, what are the methods of category confusion, and what are the characteristics of word meaning in composite categories after confusion? Thirdly, how do composition and confusion reflect the dynamic development of word meaning in composite categories, and does the dynamic categorization of composite categories involve the same processes as individual categories? These questions require further research to enhance the clarity and systematic nature of the theory of dynamic categorization.

Fourth, how the theory of dynamic categorization of word meaning can be to effectively explain common linguistic phenomena such as lexicalization, grammaticalization, and category tolerance. Liu (2009) asserted that dynamic categorization of word meaning is synonymous with grammaticalization. However, in actuality, both polysemies and the lexicalization or grammaticalization of words entail the dynamic development of semantic categories. These linguistic phenomena may merely represent different stages of this dynamic development. Once the theory of dynamic categorization is refined, it can be utilized to elucidate the relationships between these linguistic phenomena involving changes in word meaning, particularly the contentious relationship between lexicalization, grammaticalization, and subjectification.

After explaining the shortcomings in the research of dynamic categorization theory and the dynamic categorization perspective in word meaning research, we present the following diagram to envision word meaning research under the framework of dynamic categorization (Figure 1).

Figure 1:

Prospect for research on the dynamic categorization of word meaning.