Unveiling semantic complexity of the lexeme ‘reputation’: Corpus analysis

Full Text

1. Introduction

Inspired by data-driven evidence-based language studies and implying collostructional approach to language material, the current research is a corpus-based study of the lexeme ‘reputation’ and its closest lexical context. Lying within the field of collostructional semantics, the paper is aimed at unveiling the semantic complexity of the lexeme ‘reputation’ by examining its closest verbal context given in the Corpus of Contemporary American English (COCA). In the сurrent research, semantic complexity is defined as a complex semantic structure of a lexical unit characterized by a certain number of semantic components related to different aspects of the signified. Thus, the question guiding the study and defining its logic was how immediate environment enhances and contributes to the semantics of the lexeme in question.

The choice of the lexeme ‘reputation’ is predetermined by the interest in the eponymous phenomenon. There is a wide range of research, mostly in social sciences, addressing the social nature of reputation as a social construct and socially transmitted representation (Kamshilova & Chernyavskaya 2021: 54). With the growing influence of social and mass media, corporations and wealthy tech giants strive to spread their influence online by spending a substantial part of their funds on reputation management – building trust and creating positive image as visionaires, promoting themselves as pioneers of progress and bright future for humanity, fostering good publicity and maintaning their reputation – all to be used in corporate lobbying activities, negotiations with the state authorities, competition, marketing (Eslami et al. 2023: 28, Malyuga 2023: 155). Reputation is no longer seen as the opinion that is formed by the public itself but on the opposite – perceived as a valuable asset actively managed by the company and imposed on the public through the means of mass communication. This makes reputation one of the key concepts of modern public communication worth studying from a linguistic perspective. This interest is well instantiated by a series of linguistic articles that focus on academic reputation and apply the discourse analysis methodology to the data drawn from the Russian National Corpus (Chernyavskaya 2019, Kamshilova & Chernyavskaya 2021).

Dictionaries define the lexeme ‘reputation’ as “the opinion that people have about someone or something because of what has happened in the past” (LDOCE, date of reference: 29.01.2022) or very close to it “the opinion that people have about what somebody/something is like, based on what has happened in the past” (OALD, date of reference: 29.01.2022). There are three major implications in both definitions. Since it is an opinion, evaluation is part of the concept underlying reputation. Moreover, reputation is an opinion held by some people. Reputation is a consequence of what a person did in the past. All this accounts for the definition of reputation given in (Kearns et al. 2013: 3): “as a meta-belief, reputation is an evaluative belief held by an individual that s/he believes an unidentified majority to hold true about an object”. Additionally, the definition “a place in public esteem or regard: good name” given in Merriam-Webster Dictionary brings reputation closer to recognition – getting respect and being known for one’s achievements (MWE, date of reference: 31.08.2023). Such definition implies that the only form of reputation is a good reputation. With these premises in mind, we conducted the corpus-based collostructional analysis to see how language regenerates, reflects and rethinks the notion that has been in public and researchers’ eye since the 1970s and remains relevant up to this day.

To reach our goal, we found it expedient to turn to the benefits of corpus analysis as the recent growth and development of linguistic corpora and sophistication of their toolbox provide researchers with comprehensive information related to various language phenomena. “While it is possible to analyze language manually, robustness of analysis of and depth of insight into attested language use can arguably be achieved only with the aid of computational technology” (McEnery et al. 2019: 74). Corpora accelerate the search and processing of large datasets as well as form a whole new environment that requires developing specific methodologies and approaches to studying and interpreting language phenomena in the framework of corpus-based, corpus-driven and corpus-illustrated research (Dobrovol’skij 2020).

With its tools employed, corpus linguistics managed to revolutionize the way language in general and its specialized varieties are studied (Goźdź-Roszkowski 2021: 1). Thus, for example, researchers mention that by differentiating constructions and non-constructional word strings through their relative frequences, corpus analysis has given a boost to sentence parsing, which has resulted in its computerization (Brysbaert et al. 2017: 3). This achievement proves the importance of corpus analysis as the volume of linguistic data in corpora is growing yearly and manual parsing has consequently become less efficient and obsolete. It is worth mentioning that corpora have kickstarted the development and resulted in significant improvement of comparative studies (Dobrovol’skij 2020), auto-suggest algorithms and speech recognition technologies (Ulasik et al. 2020), lexicography and translatology (Durán-Muñoz & Pastor 2019). No wonder that the fast-pacing development of corpora has contributed to modern linguistics not only as a new method of research but also as a separate subject in unversity curricula (Bednarek et al. 2020: 2).

What is most important for the current study is the fact that linguistic corpora helped to prove that all languages have reliable underlying patterns that are used either by specific authors (known as idioconstructicon) or in specific works or genres (Kretzschmar 2021: 155). This perspective on a language echoes the main idea of construction grammar according to which language is seen as a network of constructions, i. e. “conventionalized form-meaning pairings” (Hoffmann 2017: 1) or “conventionalized parings of form and function” (Goldberg 2006: 4) representing basic human experiences via structures (Ramonda 2014: 67). Thе crossing point of these two trends in modern linguistics made it possible to combine these two methodologies to lay the foundation for the current study. Hence, the article is laid out in compliance with the aim and the methodology employed. The introductory part states the aim of the research and looks into the reasons for undertaking it. Further on, the theoretical background is provided and the data and methods are expatiated on. The resultative part contains the statistics and the description of all the constructions with the lexeme under analysis, followed by the discussion of the results and prospects of further studies.

2. Theoretical background

Theoretically, the research is based on the tenets of corpora analysis and collostructional approach. The use of corpus in the current research can be justified not only by its efficiency as a tool and a source of authentic language data but also by the fact that “the corpus-based approach of language analysis is more reliable as it is based on empirical data” (Shahzadi & Javed 2019: 51). Besides, the application of corpora is necessitated by the fact that “a corpus-based method can aid in explaining many issues concerning the argument structure of words and providing quantitative descriptions of their usage” (Wiliński 2021: 747). Collostructional approach helps to reveal “the lexicogrammatical associations between constructions and lexical elements” (Schmid & Küchenhof 2013: 533). The heuristic potential of collostructional approach is predetermined by its power “to predict the semantic and syntactic type of a phrase in which that word is the syntactic head” (Michaelis 2006: 73). Moreover, “construction grammarians have been very committed to identifying the function of constructions, and the delicate meaning effects that arise in context, in order to explain linguistic knowledge and language use” (Leclercq 2021: 1).

For a long time, lexis and grammar were considered separately when it came to teaching and learning English (Ruegg 2015: 1). However, in the second half of the twentieth century, the recultivation of syntactic theories led to a significant switch in studying language as a whole, embracing grammar at large and syntax in particular. Both morphology and syntax relied heavily on the notion of ‘construction’ as a formalised matrix to be filled with lexemes. Traditionally, construction was defined as a syntactic unit composed of the language entities combined in speech due to certain grammatical features (Akhmanova 2004: 202). The definition attests to the fact that phrases are heavily dependent on grammatical features of their constituents. However, despite a strong focus on grammatical combinability, definitions of construction in structural linguistics did not account for lexical and semantic combinability of their components. The situation changed in 1955 when Chomsky came up with his famous sentence “Colorless green ideas sleep furiously” and proposed its analysis, challenging commonly accepted views on syntax and phrase and thus pointing out that viable syntactic theories could not afford to ignore lexical and semantic combinability of words. The expression proved grammatically correct while being absolutely incomprehensible. In 1959, the publication of the book Elements of Structural Syntax by Tesnière brought a dramatic change to the theory of syntax. Previously, syntax had been seen as a language specific set of rules for combining words into phrases and sentences. Dependency grammar, introduced in the book by Tesnière, highlighted unequal status of the constituents in word combinations, namely, the idea that a word combination consists of the syntactic head, or the main word, with a number of valencies (dependencies) filled with mandatory (actants) and optional arguments (circumstants) (Tesnière 2015: 100–102). Words join together and form combinations not solely because they have the same grammatical features but mainly because the head predetermines and preprograms its dependent words including their grammatical features and, in later syntactic frameworks, even imposing constraints on their semantics. Such a shift in mentality did not go unnoticed for grammatical theories as well. The downwards approach from language-specific grammar rules and the necessary sets of morphemes to phrases and sentences transformed into the upwards approach – from sentences and phrases to grammar rules and sets of morphemes. Since then, word combinations have been studied as syntactic patterns rather than a set of equal lexemes with combinable grammatical features.

Thus, the necessity of addressing semantics in the study of syntax and dependency structure of word combinations recultivated syntax and gave birth to a large number of syntactic frameworks that were blurring the restricting line between lexis and grammar. Among them are Fillmore’s Case Grammar (Fillmore 1968), Minsky’s Frame-and-Slots Theory (Minsky 1974), Meaning ↔ Text Theory by Melchuk, Zholkovsky and Apresyan (Mel’čuk 1981), Halliday and Matthiessen’s Functional Grammar (Halliday & Matthiessen 2013), Langacker’s Cognitive Grammar (Langacker 1987) and Fillmore and Kay’s Construction Grammar (Fillmore & Kay 1995). Each of the frameworks is characterised by its own approach to language and linguistic phenomena as well as unique methodologies and specific key notions. The unifying feature was uncovering the connection between complex linguistic entities and their meaning.

Another great boost that linguistics experienced was inspired by the development of technologies that changed the landscape of linguistic research completely (Privalova & Kazachkova 2022, Solovyev et al. 2022). Parallel to the evolving syntactic theories and linguistic turn to semantics in the late 1950s, the first linguistic corpus to appear in the 1960s was Brown Corpus of American English followed by Lancaster-Oslo-Bergen Corpus of British English in the 1970s. The uniqueness of the corpora was that they compiled authentic texts and provided insight into linguistic data and patterns that could not be obtained by using traditional lexicographicalr esources. Further on, in the 1990s and 2000s, the relative affordability of personal computers and the Internet resulted in a surge in linguistic corpora, both synchronic and diachronic. Among them are British National Corpus, Russian National Corpus, Corpus of Contemporary American English, the Leipzig Corpora Collection and many more. Parallel corpora, such as Multext Project, Multext-East, RuN-Euro Corpus, Reverso Context, Linguee also play a vital role in modern translatology and comparative studies. Today corpora include a variety of texts of different genres, time periods, authors, and sources. Studying language through corpora led to the accumulation of evidence in favor of the mutually inseparable nature of lexis and grammar, which, in its turn, resulted in the emergence of full-fledged constructional paradigm.

At present, constructional grammar encompasses a great variety of linguistic phenomena to analyze and is constantly fueled by the fast-developing technologies for processing natural language as well as new statistical, computational, and experimental methods for studying and generalising numerous facts of language (Ackerman et al. 2014: 758). Constructional grammar is one of the most recent and actively developing spheres in linguistics. Stefanowitsch and Gries define constructional approach to language as the one that has established itself in various fields of linguistic knowledge over the years, and that poses construction as a basic lexico-syntactic sign in languages (Stefanowitsch & Gries 2003: 209). This approach, which is also in line with the ideas of lexicogrammar and the tenets of cognitive linguistics, offers a completely new perspective on language and language studies.

The cornerstone of constructional grammar is construction. The term itself was reimagined and defined as a meaningful operational unit of a language. Thus, the notion encompasses not only phrases but individual lexemes as well. It is worth mentioning though, that some areas of construction grammar, the notion of construction included, still cause heated debates. Despite disagreement, adepts of construction grammar share two things – “their love of interesting and complex data and their dislike of most work in the universal grammar camp, whose theories they regard as distorting the basic nature of individual languages to fit a pre-conceived mold” (Sag et al. 2013: 2). In other words, unlike in universal grammar, research in construction grammar stems from analysing non-predictable forms (‘many a day’, ‘all of a sudden’, ‘by and large’) and non-predictable meanings (‘break cover’, ‘show the ropes’) across massive datasets thus moving from evidence to theoretical generalizations.

Hence, the objective of the current research is achieved by means of collostructional analysis and semantic clusterisation. Semantic clusterisation helps to break the collocations into semantic clusters while collostructional analysis is aimed at defining the collocations with the strongest assosiation between components thus turning them into collostructions. Semantic cluster is an often multilayered group of collocations sharing the same resultative meaning. For example, the cluster ‘TO CREATE reputation’ comprises 14 collocations, such as ‘develop a reputation’, ‘build a reputation’, ‘cultivate a reputation’ and others which share the meaning of making a reputation. Thus, embracing the semantics of the lexeme ‘reputation’ can shed some light on its current status in the lexis, linguocultural implications as well as its functioning in the context. The underlying principle of the collostructional semantics can be formulated as follows: “you shall know a word by the company it keeps” (Firth 1957:11). In other words, the analysis of the most frequent collostructions with the lexeme ‘reputation’ as their node in COCA can help uncover its linguocultural implications and nuances of usage. No doubt, “a detailed study of the semantics of linguistic signs makes it possible to reveal the volume and hierarchy of the meanings of the word, and also to gain access to the content of concepts as units of consciousness, to reveal specific and universal moments in national world view” (Klimenko 2018: 314). Examining and interpreting the semantic network of the lexeme ‘reputation’ in ‘adjective reputation’ and ‘verb reputation’ constructions can provide an insight into its conceptualization that is defined as “a mental grasp, segmentation, specification and categorization of data pertaining to the material and abstract world and subsequently processing it in thought and language” (Bila & Ivanova 2020: 222).

3. Data and methods

The choice of constructions under study is defined by the principle of colligation which implies “linear co-occurrence preferences and restrictions holding between specific lexical items and the word-class of the items that precede or follow them” (Stefanowitsch & Gries 2003: 209). Being a noun, the lexeme ‘reputation’ can collocate with other notional parts of speech, more specifically – adjectives and verbs – which contribute to its semantics. Hence, the research is devoted to examining two types of constructions: ‘adjective reputation’ and ‘verb reputation’. In this paper, construction is a complex lexicogrammar unit which consists of the main component (the syntactic head) and at least one lexically unspecified variable. Thus, construction is a non-elementary, compound unit of language that functions as an entity (Rakhilina 2010: 19–21). At the same time, collocation represents a string of lexemes with unidentified syntactic connection and strength of association. Meanwhile, in this paper constructions with all of their slots filled with syntactically connected lexemes that demostrate statistically strong association are called collostructions (e.g. ‘professional reputation’, ‘international reputation’, ‘gain reputation’).

The research starts with forming the queries ADJ REPUTATION and VERB * REPUTATION in List section of Corpus of Contemporary American English (COCA) where the words typed in capitals represent lemmas. A lemma includes all possible grammatical variations of the lexeme in the corpus. To include the results with determiners or prepositions before ‘reputation’, the symbol * is used to introduce an additional token. Further on, the symbol (*) denotes constuctions with an additional token before ‘reputation’ and without it. The results in the queries are sorted by frequency. The query ADJ REPUTATION provides 1,325 unique forms with total frequency of 6,319 tokens while the query VERB * REPUTATION produces 2,820 unique forms and 6,888 tokens in total and VERB REPUTATION – 198 unique forms and 315 tokens. The present research encompasses top 100 entries of each of the quieries, that is 98 unique forms in ADJ REPUTATION with total frequency of 4,088 tokens and in VERB (*) REPUTATION – 57 unique forms with total frequency of 6,190. The obtained sampling is then refined by means of contextual analysis. Contextual analysis helps to clear up the initial sampling from the collocations where components have no direct syntactic connections. For example, the collocation ‘google reputation’ can be obtained as an entry of the query VERB REPUTATION. Contextual analysis provides an opportunity to look at the phrase in a wider context – ‘Just google reputation management’ – where ‘reputation’ functions as an attribute and has a primary syntactic connection to the lexeme ‘management’ in ‘noun+noun’ construction rather than with ‘google’. Another example is ‘given his reputation’ in the context ‘His dire financial straits were surprising, given his reputation as a savvy market forecaster’ where ‘given’ is a derivative preposition and not a verb.

Subsequently, the adjective and verb collocates of the lexeme ‘reputation’ are grouped into sematic clusters. Semantic clusterisation is the method of breaking collocations into groups based on the shared semantic features of their collocates (Ivanova & Medvedeva 2022: 687). A close examination of the semantic clusters is aimed at unveiling semantic complexity of the lexeme ‘reputation’ as well as identifying the current trends of usage. As a rule, clusters contain lexemes that represent the same semantic domain and are sometimes related to each other as hyponyms and hyperonyms. For example, ‘academic reputation’, ‘political reputation’ and ‘military reputation’ are hyponyms that belong to the hyperonym ‘professional reputation’. After clusterisation the resultative meanings of the collostructions in the clusters are examined and generalized.

In conclusion, the strength of association between ‘reputation’ and the verbs and adjectives in the sampling is calculated by means of log-likelihood association measure. In this research, instead of applying Fisher’s exact test to covarying collexemes as in (Stefanowitsch & Gries 2005), we choose log-likelihood measure due to the fact that Fisher’s exact test requires immense computational capacities for analyzing large datasets (Evert 2005: 80). In its turn, log-likelihood score helps rank the collostructions based on the strength of association between the lexemes in comparison to the expected frequency of their co-occurrence. The calculations used in this paper are based on the algorithm provided in (Brezina 2018: 69–72). According to S. Evert, log-likelihood score shall be interpreted as follows – the higher the score, the stronger the association (Evert 2005: 337). The null hypothesis h₀ stipulates that there is no relation between the collocates and ‘reputation’ in the sampling, hence their log-likelihood score (LL) does not exceed the expected frequency of co-occurrence (E11).

4. Results

The query ADJ REPUTATION provided 1,325 unique forms represented by 6,319 tokens. The most frequent collocations are ‘bad reputation’ (481), ‘good reputation’ (449), ‘international reputation’ (234), ‘national reputation’ (225), ‘great reputation’ (141), ‘professional reputation’ (107), ‘academic reputation’ (107), ‘online reputation’ (105). The sampling consists of 98 collocations ranged by frequency with 4,088 total number of tokens. The least frequent collocations have a frequency of 9 tokens and comprise 0,22% of the total amount of tokens in the sampling. The collocational analysis enabled to form 7 semantic clusters ranged from the most extensive to relatively small.

1. Positive adjective + REPUTATION – the cluster includes 23 unique forms represented by 1,345 tokens. Here the term ‘positive adjective’ is used to describe adjectives with positive denotative meanings. The adjectives can be grouped into several semantic sub-clusters: adjectives of size (‘growing reputation’ (77), ‘big reputation’ (28), ‘highest reputation’ (9), ‘high reputation’ (47)); adjectives of evaluation (‘impeccable reputation’ (36), ‘best reputation’ (25), ‘positive reputation’ (31), ‘better reputation’ (38), ‘good reputations’ (38), ‘good reputation’ (449), ‘excellent reputation’ (71), ‘great reputation’ (141), ‘fine reputation’ (21), ‘outstanding reputation’ (21)); adjectives of emotion and attitude (‘impressive reputation’ (13), ‘enviable reputation’ (11)); adjectives of appearance (‘spotless reputation’ (18)); adjectives of firmness and durability (‘strong reputation’ (70), ‘solid reputation’ (81)), as well as adjectives of result (‘established reputation’ (23), ‘well-earned reputation’ (48), ‘deserved reputation’ (38), ‘hard-earned reputation’ (11)). The most represented subcluster is adjectives of evaluation which includes 871 tokens (~64,8%) followed by adjectives of size – 161 tokens (~12,0%), adjectives of firmness and durability – 151 tokens (~11,2%), adjectives of result – 82 tokens (~6,1%), adjectives of emotion – 24 (~1,8%) and adjectives of appearance – 18 (~1,3%). High frequency of adjectives of evaluation substantiates the idea expressed by L.P. Poznyak that the lexeme ‘glory’ is strongly associated with the lexeme ‘reputation’ as “evaluation is a key factor in forming of axiological picture of the world” (Poznyak 2019: 8). As both ‘glory’ and ‘reputation’ belong to the same conceptual field, the factor of evaluation can be equally relevant when analysing both lexemes. Moreover, it is worth mentioning that the lexeme ‘reputation’ tends to collocate with clusters of adjectives that are typically used to describe tangible objects and their properties. Regarding this matter, Golovanivskaya mentions that abstract notions that inevitably date back to concrete notions strive to become concrete but on a completely different level – they acquire features of concrete objects through material connotation which forms secondary and eclectic concrete image attached to these abstract notions (Golovanivskaya 2018: 103). Selmistraitis and Boikova argue that “we use physical things that we have more experience with, like war, journeys, buildings, and food to understand concepts that are more abstract or actions like arguments, love, theories, and ideas. Since the majority of our experience comes from contact with the physical world, it is understandable that we will use it to comprehend abstract concepts” (Selmistraitis & Boikova 2020: 15). Thus, reputation may be characterized by a certain size and the larger the reputation, the better. It might be assumed that negative reputation cannot be large and positive reputation, on the other hand, is never small. Positive reputation glows like a precious gemstone and can be spotless like a piece of fabric. In such collocations adjective collocates are used metaphorically.

In his latest research, Patekar points out that in the recent studies on collocations published in English they are called ‘metaphorical collocations’ (Patekar 2022). The author specifies that the notion itself is rarely used, and if is, the researchers mostly consider it self-explanatory thus avoiding giving it a proper definition. Keglević Blažević also mentions that “regardless of the different approaches, it can be observed that some collocations show a kind of change in meaning. These are referred to as metaphorical collocations and are understood as a subcategory. The collocate of the metaphorical collocation has more than one meaning, and is, therefore, polysemous” (Keglević Blažević 2022: 190). Patekar proposes the following definition of metaphorical collocations: “a specific type of a collocation in which the collocate is used figuratively and the base literally, thus imbuing the collocation with metaphorical meaning and distinguishing it from a metaphorical expression in which none of the components is used literally” (Patekar 2022: 45). Consequently, this definition can be applied to the collocations in question. The collostructions where the lexeme ‘reputation’ is combined with adjectives denoting features of tangible objects have the lexeme ‘reputation’ as the base with literal meaning while adjective collocates are used figuratively.

Moreover, in COCA the form ‘N-earned reputation’, where N is a variable, can be filled with ‘well’, ‘long’ and ‘hard’, meaning that good reputation can only be gained by efforts and hard work, which is explicated on the lexical level. Meanwhile, the expression ‘easy-earned reputation’ does not occur in COCA. Search via Google provides only four relevant entries. Among them there is a message on the text-based RPG forum that warns “when factions with 700,000 and 1,400,000 respect, with membership requirements like a minimum of 1,000,000 in battle stats per player, declare war on a new faction with mostly level 3–5 member, with combined total battle stats of all members of not more than maybe 300,000... is it fair to say those faction leaders are nothing but absolute cowards? Looking to make some easy-earned reputation by beating on two-week-old players. What a bunch of losers” (TORN, date of reference 15.08.2022). Another example can be found on the website of Nigerian daily newspaper, Blueprint Newspaper: “Tax Collectors have had a long easy-earned reputation of skewing the process. The famed short, mean Zacchaeus in the Holy Book often cheated on people” (Blueprint, date of reference: 15.08.2022).

It can be concluded that the expression ‘easy-earned reputation’, however possible, denotes the reputation gained in a dishonest way by cheating, abuse of power or due to unfair competition. Such reputation is widely reproached and is considered a shame. The examples in question prove that social and cultural expectations and values are embraced in language and going against the social norms and conventions often results in a public disgrace. This substantiates the idea that “metaphor is part of the system of human thinking that conceptualizes one concept to another in the form of life behavior as a sociocultural and historical experience of a society” (Sarif et al. 2020: 54). This stance is supported by Kozlova positing that metaphor most fully reflects ethnic and cultural relatedness of cognition as it embodies culture-specific experience (Kozlova 2020: 919). It can be further assumed that the balance between socially accepted and socially disapproved actions can be reflected in the frequency of such collocations.

In terms of genres, the majority of the collocations (~63%) can be found in newspaper (231), magazine (207), blog (205) and web-general (182) sections of COCA. The spoken section is represented by 164 collocations (~12,5%) from this cluster.

2. Negative adjective + REPUTATION – the cluster includes 17 unique forms represented by 886 tokens. As in the cluster above, the term ‘negative adjective’ is used to describe adjectives with negative denotative meaning correspondingly. The negative adjectives can be further grouped into the following semantic sub-clusters: adjectives of evaluation (‘bad reputation’ (481), ‘bad reputations’ (24), ‘negative reputation’ (37), ‘the worst reputation’ (18), ‘the worst reputations’ (10), ‘nasty reputation’ (18), ‘poor reputation’ (48)); adjectives of emotion (‘fearsome reputation’ (36), ‘terrible reputation’ (31), ‘horrible reputation’ (19)); adjectives of physical condition (‘tarnished reputation’ (51), ‘damaged reputation’ (18), ‘tattered reputation’ (11), ‘battered reputation’ (11)) and adjectives of attitude (‘undeserved reputation’ (17), ‘notorious reputation’ (29), ‘unsavory reputation’ (27)). Additionally, beyond the imposed limitations on frequency, we can find adjectives of pattern and texture – ‘checkered’ (7) and ‘fraying’ (1). The adjective ‘checkered’ falls into the cluster because it does not only define a pattern but also has the metaphorical meaning in collocations with abstract nouns such as ‘past’, ‘reputation’, ‘career’ – “marked by alternation or contrast of fortune; marked by many problems or failures” (MWD, date of reference: 15.08.2023). The adjective ‘fraying’ is derived from the verb ‘fray’ that means “to become or to cause the threads in cloth or rope to become slightly separated, forming loose threads at the edge or end” (CDO, date of reference: 15.08.2023). The fact that this verb collocates with the lexeme ‘reputation’ substantiates the idea that reputation is metaphorically represented as a piece of fabric.

The subcluster with adjectives of evaluation comprises almost 636 tokens (~71,8%), adjectives of emotion reach the number of 86 and represent ~9,7% of the cluster, while the subcluster of adjectives denoting physical condition have 91 tokens per subcluster (~10,3%) and adjectives of attitude – 73 tokens (~8,2%). The cluster shows the same tendency as the previous one: the lexeme ‘reputation’ collocates with the adjectives denoting properties of tangible objects. The current cluster also lacks the subcluster with adjectives of size. However, a case of metonymy in the collocations ‘notorious reputation’, ‘unfortunate reputation’ and ‘esteemed reputation’ in which the mentioned qualities of a person are attributed to the reputation is also worth observing (LDOCE, date of reference: 29.01.2022). The attribution of the quality typical for a human being enables to personificate reputation thus creating an additional metaphorical representation – ‘reputation as a human being’. Kuznetsova argues that while in metaphor the source domain is mapped onto the target domain, metonymy is based on establishing connection between elements of the same conceptual structure (Kuznetsova 2021: 73). Thus, it can be concluded that metaphor can merge with metonymy. The metaphorical representation of reputation as a human being helps to depict reputation, especially bad reputation, as something independent from its proprietor, something that lives its own life, something that cannot be controlled. Furthermore, the majority of collocations are found in newspaper (129), magazine (128), web-general (119) and blog (138) sections of COCA (~58%).

3. Adjectives of scope and environment + REPUTATION – the cluster is represented by 14 unique forms and 814 tokens. The collocations from this cluster can be alternated as follows ‘regional reputation’ – ‘reputation in the region’, ‘public reputation’ – ‘reputation with the public’, ‘critical reputation’ – ‘reputation with the critics’. The shared seme of the adjectives is ‘a place, territory or a community’ as an environment for reputation. It is worth mentioning that in collocation ‘historical reputation’ (16) history is perceived as an environment for reputation. The most frequent collocations in the cluster are ‘international reputation’ (234), ‘national reputation’ (225) and ‘online reputation’ (102), with the collocation ‘international reputation’ being used the most in academic (60), newspaper (43) and magazine (42) sections, ‘national reputation’ in newspaper section (99) and ‘online reputation’ in blog section (64) of COCA. It can be assumed that in American culture, the wide use of the collocation ‘national reputation’ in the press has a stronger appeal.
4. Adjectives of status + REPUTATION – the cluster comprises 23 unique forms and 637 tokens. The collocations from this cluster can be alternated as follows: ‘artistic reputation’ – ‘reputation as an artist’, ‘progressive reputation’ – ‘the reputation of being progressive’. In the cluster three subclusters can be identified. The first subcluster includes adjectives denoting fields of activities (‘political reputation’ (20), ‘artistic reputation’ (18), ‘literary reputation’ (44), ‘academic reputation’ (105), ‘intellectual reputation’ (9), ‘literary reputations’ (10), ‘academic reputations’ (9), ‘military reputations’ (10), ‘scholarly reputation’ (11), ‘scientific reputation’ (25), ‘defensive reputation’ (10)). Collocations with the adjectives of position and status relate to the second subcluster: ‘institutional reputation’ (10), ‘professional reputation’ (107), ‘professional reputations’ (24), ‘corporate reputation’ (36), ‘personal reputation’ (48), ‘corporate reputations’ (11), ‘personal reputations’ (18), ‘stellar reputation’ (65), ‘legendary reputation’ (11). The third cluster consists of collocations with adjectives denoting personal outlooks – ‘conservative reputation’ (12), ‘liberal reputation’ (11), ‘progressive reputation’ (13). Thus, collocations with adjectives of fields of acitivites account for 42,5%, collocations with adjectives of status – 52%, collocations with adjectives of outlook – 5,5%. From the stylistic standpoint, the majority of the collocations in the cluster are used in academic (154), newspaper (102) and magazine (123) sections of COCA.
5. Adjectives of uncertainty + REPUTATION – the cluster includes 8 unique forms and 210 tokens. Adjectives in the cluster can denote either positive or negative reputation depending on the general context, for instance, ‘certain reputation’ (23), ‘controversial reputation’ (9), ‘dubious reputation’ (18), ‘mixed reputation’ (16). In the collocations ‘considerable reputation’ (27) and ‘formidable reputation’ (16), the adjectives make the lexeme ‘reputation’ function as a synonym of ‘power’ and ‘influence’. Most collocations from the cluster belong to magazine (45), newspaper (41) and fiction (28) sections of COCA.
6. Adjective of time + REPUTATION – the cluster comprises 9 unique forms and 133 tokens. The collocations from the cluster contain adjectives which denote age, time period and longevity of reputation – ‘enduring reputation’ (9), ‘new reputation’ (22), ‘longstanding reputation’ (18), ‘posthumous reputation’ (16), ‘past reputation’ (15), ‘old reputation’ (12), ‘long-standing reputation’ (15), ‘early reputation’ (15), ‘prior reputation’ (11). The collocations from the cluster are the most frequent in academic (38), magazine (37) and newspaper (25) sections of COCA.
7. Adjectives of wholesomeness + REPUTATION – the cluster consists of 4 unique forms and 63 tokens. Collocations ‘general reputation’ (23) and ‘overall reputation’ (14) are used in academic, blog and web-general sections while ‘entire reputation’ (11) and ‘whole reputation’ (15) in movies, magazines and spoken sections of COCA.

Moreover, in the selection there are 199 cases of using the lexeme ‘reputation’, which is an abstract noun, in its plural form. This can be explained by the examination of the immediate context: the lexeme ‘reputations’ occurs when talking about a number of entities (companies, people, institutions, organizations, etc.) and thus underlining an individual character character of each of their reputations.

Speaking of the metaphorical collocations with the lexeme ‘reputation’, two major types of representation can be identified – reputation as a tangible object (objectification based on metaphor) and reputation as a person (personification based on metonymy). According to Šeškauskiené and Stepančuk, such tendencies “are in line with the cognitive principle of embodiment, because our perception of abstractions in terms of objects or humans arises from our interaction with the world, where people and the material world taking the form of concrete objects are the main ‘interacting sides’” (Šeškauskiené & Stepančuk 2014: 116).

The query VERB * REPUTATION provided 2,354 unique forms represented by 6,886 tokens. The initial sampling includes the first one hundred of collocations ranged by frequency. After excluding collocations with modal verbs and auxiliary verbs from the selection, 57 unique forms with 6,190 total number of tokens have been obtained. The collocations are further grouped into eight semantic clusters and sorted by frequency.

1. ‘TO GET’ verbs + REPUTATION – the cluster is the largest in the query results and is represented by 8 main verbs and 2,015 collocations. The most frequent verbs in the cluster are ‘earn’ – 760 tokens (~37,7%), ‘get’ – 550 (~27,3%), ‘gain’ – 423 (~20,1%). The collocations are frequently used in magazine (443), newspaper (439), web-general (232) sections of COCA (~55,3%). The collocations are the least represented in fiction (149), blog (180) and TV/movies (182) sections. It is worth mentioning, the lexeme ‘reputation’ collocates with the verb ‘earn’, which means that reputation can be obtained as a result of work and efforts, and the verb ‘gain’, which, as the previous research suggests, “is most frequently used to point to a progressive step-by-step change, a steady improvement specifically related to intellectual abilities, skills, power and control” (Ivanova & Medvedeva 2022: 689).
2. ‘TO CREATE’ verbs + REPUTATION – the cluster includes 16 unique verb forms and 1,992 tokens. The most frequent verbs in the cluster are ‘build’ – 686 (~34,4%) and ‘develop’ – 407 (~20,4%). The cluster also includes the verb ‘burnish’ with its literary meaning ‘to rub metal until it is smooth and shiny’, the verb ‘forge’ – “to form (something, such as metal) by heating and hammering; to form or bring into being especially by an expenditure of effort” as well as the verb ‘cultivate’ – ‘to prepare land and grow crops on it, or to grow a particular crop’, the verb ‘garner’ – ‘to collect something, usually after much work or with difficulty’, the verb ‘bolster’ – ‘to support or improve something or make it stronger’ and the verb ‘cement’ – ‘to put cement on a surface or stick things together using cement’(CDO, date of reference: 29.01.2022; MWD, date of reference 15.08.2023). The mentioned verbs denote manipulations with a tangible object and point out to metaphorical representation of reputation. The actions represented by these verbs imply intensions and efforts put into improving the properties of a tangible object. The properties to be improved correlate with those in ‘positive adjective + REPUTATION’ cluster – adjectives of size (‘big reputation’, ‘wide reputation’), appearance (‘glowing reputation’) as well as firmness and durability (‘solid reputation’, ‘strong reputation’). The collocations have a high frequency rate in newspaper (513) and magazine (427) sections of COCA.
3. ‘TO DESTROY’ verbs + REPUTATION – the cluster comprises 17 unique verb forms and 1,194 tokens. The most frequent verbs are ‘ruin’ – 274, ‘damage’ – 263 and ‘destroy’ – 190, which mirror the verbs ‘build’ and ‘cement’ from the previous cluster. Additionally, the current cluster includes the verbs ‘sully’ – ‘to spoil something that is pure’ and ‘tarnish’ – ‘to make or (especially of metal) become less bright or a different color’ (CDO, date of reference: 29.01.2022). The mentioned literal meanings of the verbs again correlate with such qualities of a good reputation as being spotless and glowing. Moreover, co-occurence with the verb ‘kill’ personifies reputation.The collocations from this cluster fall into the category where, according to Vinogradova and Vorobyova, imagery and value components are realised by the semantically close lexemes denoting authority, respect, grace, fame and approval, on the one hand, combined with the verbs which explicate the semes of loss, depreciation and damage done to the social status of an individual, on the other hand (Vinogradova & Vorobyova 2019: 148). The collocations from the cluster tend to fall into blog (233), web-general (176) and spoken (175) sections of COCA.
4. ‘TO SAVE’ verbs + REPUTATION – the cluster consists of 296 tokens and 4 verbs: protect – 180, save – 54, defend – 31, salvage – 31. Interestingly enough, the literary meaning of the verb ‘salvage’ is ‘to save goods from damage or destruction, especially from a ship that has sunk or been damaged or a building that has been damaged by fire or a flood’ (CDO, date of reference: 29.01.2022) denotes manipulation with a tangible object, efforts put into keeping reputation from damage and destruction.
5. ‘TO RESTORE’ verbs + REPUTATION – the cluster comprises 189 tokens and 5 unique verb forms: ‘restore’ – 96, ‘repair’ – 28, ‘rebuild’ – 23, ‘redeem’ – 23, ‘rehabilitate’ – 19. The majority of collocations are used in newspaper (43) and spoken (35) sections of COCA. The verb ‘rehabilitate’ personifies reputation as it means ‘to return someone to a good, healthy, or normal life or condition after they have been in prison, been very ill, etc.’ or together with the verbs ‘rebuild’, ‘restore’ and ‘repair’ contributes to metaphorical representation ‘reputation is a tangible object’ with its meaning ‘to return something to a good condition’ (CDO, date of reference: 18.08.2023).
6. ‘TO CARE FOR’ verbs + REPUTATION – the cluster includes 177 tokens and 4 unique verb forms: ‘maintain’ – 77, ‘keep’ – 38, ‘preserve’ – 33, ‘care’ – 29. Among the verbs, the verb ‘maintain’ has a literary meaning ‘not allow to become less’ (CDO, date of reference: 29.01.2022) which substantiates that the larger the reputation, the better. It can be assumed that reputation can be damaged not only by actions but also by time itself and thus needs to be taken care of. The key sections for the collocations in COCA are blog (33), web-general (31), newspaper (27) and magazine (26).
7. ‘TO RISK’ verbs + REPUTATION – the cluster numbers 170 tokens and 2 unique verb forms: ‘stake’ – 97 and ‘risk’ – 73. Reputation is again perceived as a tangible and valuable object that can be put as a stake in gambling. The collocations are frequently used in TV/movies (35), newspaper (29) and blog (27) sections.
8. TO ENJOY + REPUTATION – the cluster is represented by the verb ‘enjoy’ accounting for 157 tokens. This specific cluster substantiates that having a reputation can bring positive emotions and the joy of one’s accomplishments. The collocations from the cluster belong to magazine (46), newspaper (31) and academic (28) sections.

Beyond the sampling analysed in the current research, an additional cluster can be obtained. Generally, it encompasses verbs of evaluation such as ‘gauge’, ‘monitor’ and ‘evaluate’ that rarely collocate with ‘reputation’.

Further on, the table below represents adjectives and verbs as collocates of ‘reputation’ ranked from high to low log-likelihood score. ‘Obs. Freq., O11’ stands for ‘observed frequency of co-ocurrence’ and corresponds to the number of exact hits in the corpus (‘good reputation’ – 449, ‘international reputation’ – 234) while ‘Exp. Freq., E11’ stands for ‘expected frequency of co-occurrence’ – the chance of random co-occurence of the linguistic variables in question.

Table 1. Log-likelihood score of collocates in ‘adjective reputation’ and ‘verb * reputation’ constructions

ADJ+REPUTATION				VERB * REPUTATION
Collocate	Obs. Freq., O11	Exp. Freq., E11	Log-likelihood score, LL	Collocate	Obs. Freq., O11	Exp. Freq., E11	Log-likelihood score, LL
bad	481	4,15437	1575,25	earn	760	3,95312	2825,60
good	449	31,08279	681,23	build	686	12,86475	1787,16
international	234	4,40988	608,75	gain	423	3,92953	1357,22
national	225	9,00159	442,21	ruin	274	1,34831	1029,87
well-earned	48	0,00749	327,76	develop	407	9,44395	983,96
tarnished	51	0,01703	312,26	damage	263	1,38561	972,84
academic	105	1,61389	291,24	tarnish	105	0,10434	542,20
stellar	65	0,14835	287,49	destroy	190	4,31048	462,39
online	102	1,64612	278,66	stake	97	0,26516	414,15
professional	107	2,26199	267,68	protect	180	7,08321	353,14
solid	81	1,12694	231,56	hurt	54	0,02035	326,58
deserved	38	0,01647	223,72	enjoy	157	7,24537	286,68
growing	77	1,47236	199,18	risk	97	1,37129	275,48
excellent	71	1,30979	185,82	acquire	102	2,01825	260,10
fearsome	36	0,04250	179,97	restore	96	1,60899	258,53
impeccable	36	0,05729	170,55	besmirch	93	1,70377	243,31
great	141	15,43287	162,13	cement	55	0,14337	237,08
unsavory	27	0,03101	135,63	sully	33	0,09802	138,46
literary	44	0,65159	123,41	enhance	86	5,99771	126,96
strong	70	4,19731	114,02	get	550	187,18663	121,75
notorious	29	0,19808	100,65	burnish	25	0,03741	119,97
undeserved	17	0,01493	89,41	maintain	77	5,07369	117,33
professional	24	0,15755	84,19	salvage	31	0,20868	107,95
spotless	18	0,03545	81,92	bolster	30	0,34383	90,62
posthumous	16	0,02396	76,65	cultivate	31	0,47786	85,69
poor	48	3,39206	71,77	redeem	23	0,26421	69,42
corporate	36	1,44577	70,54	rehabilitate	19	0,12782	66,17
longstanding	18	0,07912	69,34	repair	28	0,82773	61,71
established	23	0,27064	69,04	preserve	33	1,77261	55,95
considerable	27	0,56048	67,93	rebuild	23	0,90898	44,98
negative	37	1,80632	66,50	forge	19	0,51404	43,32
damaged	18	0,12584	62,11	save	54	9,02877	40,95
dubious	18	0,14211	60,22	garner	13	0,18083	37,09
personal	48	4,59328	60,17	defend	31	3,12153	36,26
terrible	31	1,45822	56,67	trash	14	0,29641	34,87
outstanding	21	0,43444	52,90	affect	26	5,48383	14,96
long-standing	15	0,10700	51,49	gauge	1	0,05650	1,68
formidable	16	0,16708	49,67	monitor	2	1,76042	0,75
enviable	11	0,02438	48,92	care	29	12,77033	0,82
hard-earned	11	0,03042	46,80	evaluate	2	1,98934	0,86
historical	27	1,53862	45,09	kill	17	17,18339	7,46
nasty	18	0,43693	42,89	keep	38	35,03091	15,10
positive	31	2,57898	42,28	–	–	–	–
artistic	18	0,47132	41,74	–	–	–	–
tattered	11	0,06416	39,68	–	–	–	–
scientific	25	1,60891	39,27	–	–	–	–
mixed	16	0,36435	38,99	–	–	–	–
literary	10	0,04539	38,24	–	–	–	–
horrible	19	0,76762	37,13	–	–	–	–
battered	11	0,12595	33,27	–	–	–	–
worst	10	0,12062	29,81	–	–	–	–
better	38	6,91876	29,24	–	–	–	–
high	47	10,70497	28,89	–	–	–	–
checkered	7	0,02673	27,82	–	–	–	–
academic	9	0,11241	26,56	–	–	–	–
scholarly	10	0,18894	25,96	–	–	–	–
legendary	11	0,29689	25,22	–	–	–	–
enduring	9	0,15959	23,85	–	–	–	–
impressive	13	0,66227	22,90	–	–	–	–
worst	18	1,73176	22,48	–	–	–	–
military	10	0,31853	21,55	–	–	–	–
glowing	8	0,13901	21,34	–	–	–	–
prior	11	0,46061	21,17	–	–	–	–
unfortunate	10	0,37126	20,25	–	–	–	–
progressive	11	0,55125	19,53	–	–	–	–
overall	14	1,21305	18,64	–	–	–	–
general	23	4,09040	18,08	–	–	–	–
institutional	10	0,51661	17,50	–	–	–	–
certain	23	4,30727	17,24	–	–	–	–
controversial	9	0,58329	14,08	–	–	–	–
defensive	10	0,79558	13,99	–	–	–	–
conservative	12	1,37406	13,36	–	–	–	–
intellectual	9	0,70656	12,69	–	–	–	–
fine	21	4,97547	12,35	–	–	–	–
liberal	11	1,24859	12,32	–	–	–	–
past	15	3,60233	8,69	–	–	–	–
highest	9	1,27042	8,59	–	–	–	–
wide	10	1,84457	7,60	–	–	–	–
best	25	9,72976	7,23	–	–	–	–
political	20	7,81211	5,75	–	–	–	–
early	15	4,98928	5,65	–	–	–	–
big	28	13,14875	5,49	–	–	–	–
entire	11	3,42211	4,58	–	–	–	–
fraying	1	0,00341	4,07	–	–	–	–
whole	15	6,51890	3,49	–	–	–	–
old	12	6,78736	1,41	–	–	–	–
new	22	27,96813	0,60	–	–	–	–

The null hypothesis h₀ stipulating non-existent association between ‘reputation’ and its collocates proved true in a few cases with adjectives ‘best reputation’ (LL 7,23 < E11 9,73), ‘political reputation’ (LL 5,75 < E11 7,81), ‘big reputation’ (LL 5,49 < E11 13,15), ‘whole reputation’ (LL 3,49 < E11 6,52), ‘old reputation’ (LL 1,41 < E11 6,79), ‘new reputation’ (LL 0,60 < E11 27,98) as well as verbs – ‘monitor reputation’ (LL 0,75 < E11 1,76), ‘care * reputation’ (LL 0,82 < E11 12,77), ‘evaluate reputation’ (LL 0,86 < E11 1,99), ‘kill reputation’ (LL 7,46 < E11 17,18), ‘keep reputation’ (LL 15,1 < E11 35,03) and even ‘get reputation’ (LL 121,75 < E11 181,19). Despite relatively high frequency of co-occurrence in COCA, these collocations are rather occasional. On the one hand, weak association can be explained by semantics of some of the collocates, for example, reputation can not be broken into separate pieces. However, ‘whole reputation’ implies otherwise. Reputation can not be old or new because once established it can not be replaced with a new one despite the fact that reputation is metaphorically represented as a tangible valuable object. On the other hand, high expected frequency of such collocations can be explained by the fact that the collocates in question have a very high observed frequency in COCA on their own and “prefer” other nouns.

As reputation invokes emotion, adjective collocates of the lexeme are more likely to denote attitude and evaluation rather than the size of it – ‘excellent’, ‘good’, ‘bad’, ‘nasty’, ‘impeccable’. As is seen fron the chart, adjectives of evaluation, emotion and attitude have a stronger association with ‘reputation’. The strongest association can be observed in collostructions with adjectives that express evaluation (‘bad’, ‘good’, ‘solid’, ‘tarnished’, ‘stellar’), scope and environment (‘national’, ‘international’, ‘academic’, ‘online’), field of activities (‘professional’) and result (‘well-earned’, ‘deserved’). Adjective ‘high’ helps introduce metaphorical representation of a scale making it possible to measure reputation. As for the verbs most strongly associated with the lexeme ‘reputation’, it is worth mentioning that they denote actions that imply effort, time, hard work and bring positive results – ‘earn’, ‘build’, ‘develop’, ‘gain’ – while ‘ruin’ and ‘damage’ represent uncautious and careless actions with unwanted consequences. Thus, it can be assumed that being a fragile object, reputation requires active and constant efforts to make it solid and caution so as not to damage it.

5. Discussion

The results of the current research justify the statement made by Golovanivskaya about abstract notions striving to become concrete by acquiring features of tangible objects through material connotation which forms secondary and eclectic concrete image attached to these abstract notions. This is exemplified by collostructions with adjectives of condition (‘tarnished’, ‘tattered’, ‘battered’), appearance (‘spotless’, ‘glowing’) as well as adjectives of pattern and texture (‘checkered’, ‘fraying’) where reputation is metaphorically represented as a piece of fabric or as a valuable object made of precious metal. These metaphorical representations might explain why ‘reputation’ collocates with adjectives of firmness and durability such as ‘strong’ and ‘solid’. Additionally, the adjectives of appearance – ‘glowing’ and ‘tarnished’ – indicate that reputation is metaphorically related to visibility (Anderson & Shirako 2008: 320). Meanwhile, verbs ‘build’, ‘cement’, ‘destroy’, ‘rebuild’, ‘repair’ activate metaphorical representation ‘reputation as a building’ and verbs ‘tarnish’, ‘forge’, ‘burnish’ help reperesent reputation as an object made of precious metal. Moreover, the verb ‘salvage’ indicates that reputation is of utmost importance. This substantiates the idea expressed in (Sarif et al. 2020: 54) who claim that metaphor serves as an essential component of human thinking that conceptualizes sociocultural and historical experience of a society in the form of behavior.

Chernyavskaya also points out that reputation serves as “seals of approval or disapproval” and requires taking into consideration what is assumed to be “positive, desirable and obligatory by the representative majority” (Chernyavskaya 2022: 65). This statement is substantiated by two groups of adjective collocates that express approval and disapproval – ‘good’, ‘positive’, ‘excellent’, ‘impeccable’, ‘outstanding’ and ‘bad’, ‘nasty’, ‘negative’. The social expectations, values and beliefs are to some extent reflected by the collocation ‘ADV-earned reputation’ that can be filled with ‘well’, ‘long’ and ‘hard’. This means that only hard work, efforts and time can yield positive results and fame. In contrast, ‘easy-earned reputation’ is reproached because of having been obtained in a dishonest way by cheating, abuse of power or due to unfair competition.

Additionally, the research elaborates on the definition given in (Kearns et al. 2013: 3) that reputation is an evaluative meta-belief resulted from a person’s actions and behavior in the past. The ‘meta-belief’ nature of reputation is exemplified by adjectives that denote fields of activities – ‘literary’, ‘scholarly’, ‘academic’, ‘military’ etc. These fileds metonymically represent members of professional communities who evaluate their colleague – the proprietor of reputation. When defining reputation, Chernyavskaya states that “reputation is an obtained and long-standing public appraisal” (Chernyavskaya 2022: 65). The results of the current research testify that the adjectives of time that collocate with ‘reputation’ usually denote a long period of time that began in the past, for example, ‘early’, ‘past’, ‘prior’, ‘enduring’, ‘long-standing’.

6. Conclusions

Overall, 7 clusters of ‘adjective reputation’ and 8 clusters of ‘verb reputation’ have been obtained. The collocations with the lexeme ‘reputation’ as their main component are mostly found in newspaper, magazine, blog and web-general sections of COCA. On the collocational level, reputation has a wide metaphorical representation both in ‘adjective reputation’ and ‘verb reputation’ constructions where it inherits various properties of tangible objects.

The most typical metaphorical representations of reputation are ‘reputation as a building’, ‘reputation as a piece of fabric’, ‘reputation as a valuable object made of precious metal’. Just as tangible objects, positive reputation, unlike the negative one, can have a certain size, and the bigger the reputation, the better it is. Positive reputation is always solid and strong, it can be earned as a reward, built, burnished, salvaged or destroyed. Positive reputation is spotless, pure and glowing. Once gained, reputation should be maintained, cemented and protected and, if damaged or destroyed, needs repairing and rebuilding. Reputation can provoke emotions, such as envy and fear, and can be used to influence others. Reputation is not gained easily, though it can be undeserved. Gaining reputation requires efforts, hard work and patience, whereas easy-earned reputation is considered to be a disgrace.

It has been established that the adjectives of evaluation, scope and environment, field of activities, emotion and attitude have the strongest association with ‘reputation’. However, despite strong material connotations, reputation is very unlikely to be old or new because it is perceived as an inherent part of a person and can not be thrown away or replaced. Meanwhile, the verbs that show the strongest association with ‘reputation’ denote actions that imply effort, time and hard work, on the one hand, or uncautious and careless actions with unwanted consequences, on the other hand. Such verbs help represent reputation as a fragile object that requires constant attention and efforts to make and keep it solid as well as caution so as not to break it.

Despite being an abstract noun, reputation can be used as a countable noun meaning that reputation is something deeply individual. Through metonymy reputation can have the same qualities as its proprietor (‘notorious reputation’, ‘unfortunate reputation’). Moreover, it can be assumed that the cases of metonymy ‘notorious reputation’, ‘unfortunate reputation’, ‘reputations’ help personify this notion. Unlike its proprietor, reputation is omnipresent – it precedes and substitutes the person when s/he is not around. In other words, “reputation also works as an “information strainer”, which reflects evaluative attitude to a personality or an institution” (Chernyavskaya 2022: 65).

Reputation accumulates over time, exists in social environment and changes throughout life. Such strong emphasis on time helps predict the future actions and behavior of a person thus giving control and minimizing uncertainty due to its evaluative and categorizing nature.

Thus, the outcomes of the collostructional analysis make it possible to outline the semantic complexity of the lexeme ‘reputation’ by examining its linguistic habitat. As an abstract noun, the lexeme ‘reputation’ has an extensive and varied representation in metaphorical collocations. The obtained data may be used in second language teaching and learning, studies on metaphorical processing and conceptualization as well as cultural linguistics, lexicography and translatology. The research may be further extended with results of corpus-based and corpus-driven analysis of the lexemes ‘image’, ‘face’, ‘fame’ and ‘recognition’, which sometimes can be used interchangeably, in ‘adjective noun’ and ‘verb noun’ constructions.

About the authors

Svetlana V. Ivanova

St. Petersburg State University

Author for correspondence.
Email: svet_victoria@mail.ru
ORCID iD: 0000-0002-0127-9934

Dr Habil. in Philology, Professor of St. Petersburg State University. Her research interests include cultural linguistics, text linguistics, English grammar, discourse analysis, and media linguistics. She has over 240 publications in Russian and English including monographs, book chapters and articles in peer-reviewed journals.

St. Petersburg, Russia

Svetlana N. Medvedeva

Saint-Petersburg State University of Economics

Email: medvedeva.s@unecon.ru
ORCID iD: 0000-0002-6694-8385

Assistant Professor at Saint-Petersburg State University of Economics. Her research interests and publications deal with collostructional semantics, corpus studies, construction grammar and near synonyms. She has nearly 20 publications in Russian and English which include articles in peer-reviewed journals such as Lecture Notes in Networks and Systems and Cognitive Studies of Language.

St. Petersburg, Russia

References