Metonymy as a semiotic resource in fictional narrative

1 Introduction

Compared with metaphor, which has been laboriously studied in cognitive linguistics since the 1980s, and in cognitive stylistics (Semino and Culpeper 2002) and cognitive poetics (Gavins and Steen 2003; Stockwell 2002, 2019 in the last two decades, metonymy has been eclipsed and has hardly attracted attention from practitioners, especially in textual and discourse studies. Cognitive linguistics has challenged the traditional definitions of metonymy (i.e. that it operates on the names of things; that it substitutes the name of one thing for that of another; and that the relationship between two entities is often located in the world of reality) by making three major assumptions in the period from Lakoff and Johnson (1980) until Littlemore (2015): that (1) it is a conceptual phenomenon as well as a linguistic matter; (2) it is a cognitive process; and (3) it operates within an idealized cognitive model. However, little has been discussed about metonymy as a semiotic resource for creative meaning-making.

Lakoff and Johnson (1980: 35) originally discuss metonymy as a related but distinct case from “personification metaphor,” which serves the function of “providing understanding” and has “primarily a referential function” (1980: 36). This point echoes Nunberg’s (1978, 1979 observation of metonymy as “deferred reference,” in which a speaker uses a description of a and succeeds in referring to b, on the basis of a focus on “property transfer” (Nunberg 1995: 125). Rochester and Martin (1977) take referring as “art,” i.e. “the speaker’s ability to guide listeners to select precise referents of noun phrases.” Focusing on the referential function of metonymy, this article will take as background Papafragou’s (1996) observation of metonymy as “referential shorthand” and Littlemore’s (2015) observation of metonymy as “hidden shortcuts” in language, thought, and communication, and propose that metonymy is a device of producing “cognitive roundabouts” in literary creation. In analyzing one of O. Henry’s best-known short stories, “The Cop and the Anthem,” written in December 1904, this study has found that metonymy and chain metonymies actualize pragmatic and discursive functions of estranging or defamiliarizing fictional referents in Shklovsky’s (1965[1917]) sense. In metonymically encoding the characters, settings, actions, and sequence of events in extraordinarily roundabout ways, O. Henry comically presents an estranged but more vivid fictional world in a text that requires more cognitive effort of the reader to process.

2 Metonymy as a semiotic resource

Metonymy was distinguished from “personification metaphors” by Lakoff and Johnson (1980: 35): while a term such as “inflation” in “Inflation robbed me of my savings” does not refer to a person but only imputes human qualities to things that are not human, a term such as “ham sandwich” in “The ham sandwich is waiting for his check” only refers to an actual person by “using one entity to refer to another that is related to it.” Considering that both terms are inanimate and are the agents of the actions of “rob” and “wait,” or Actors of the Material Processes in the Transitivity system developed by Halliday (1985) and Halliday and Matthiessen (2014) in functional linguistics, Lakoff and Johnson’s distinction misses at least one point. Neither inflation nor a ham sandwich can carry out an action process; we only conceptualize that they do and put them as agents in the syntactic structure (or Actors in the action process). One concretizes the abstract force in the sluggish economic situation and the other distinguishes the customer with his order of a ham sandwich; both terms present more vivid conceptualizations than otherwise, except for a difference in terms of humanization or dehumanization. Hence, personification metaphors are first of all metonymic, and the key to characterizing metonymy does not squarely lie in whether we are imputing human qualities to things, but rather, in how the substituted term leads us along the cognitive paths to figure out the intended referent in terms of their logical relations.

Lakoff and Johnson (1980: 38–39) have listed seven types of metonymic concepts:

1. part for whole: The Giants need a stronger arm in right field.

2. producer for product: I hate to read Heidegger.

3. object used for user: We need a better glove at third base.

4. controller for controlled: Napoleon lost at Waterloo.

5. institution for people responsible: The Senate thinks it is immoral.

6. place for institution: The White House isn’t saying anything.

7. place for event: Watergate changed our politics.

These types of metonymic concepts, which instantiate “general princples” that do not just “occur one by one” (Lakoff 1987: 77), consistently center on references of one or another. Descriptively and explanatorily adequate as they may seem, the entities involved are far from being complete and the paths for realizing the metonymies are more complex than they may seem. This is because entities such as time, location, participants, objects, events, cause, effect, etc. all can be referred to in some other terms; and metonymies can be realized in multiple processes instead of a single substitution. For instance, while a “glove” in the baseball game in example (3) can be codified as object used for user, the substitution can be conceptualized on a number of logical and cognitive relationships such as object for function (the glove for the baseball catcher), possessed for possessor (the hand that wears the glove), and part for whole (hand for the body of the player); while “Napoleon” in example (4) is controller for controlled, the substitution can be conceptualized on logical relationships such as individual for collective, high for low, and part for whole, activating a series of mental associations like the officer standing for rank and file, the general standing for troops, and the leader standing for the led, etc. The examples given for (5) and (6) are basically the same in that they both refer to people by way of their location or affiliation: “Senate” is a part for whole metonymy derived from “The United States Senate” and an institution for people metonymy at the same time; and “White House” is not exactly a place for institution metonymy, since it refers to an official or a spokesperson who is not physically located there but symbolically represents the US government headed by the President who lives in the official home at “1600 Pennsylvania Ave, NW Washington, DC 20500.” Thus, it is a metonymy realized by location for located, material for symbolic, premises for institution, and institution for people substitutions. The example given for (7) is not simply a place for event metonymy, either. Instead, it is first of all a part for whole lexical metonymy (i.e. standing for an encyclopedic term “Watergate Scandal”) rather than using the place for the event; and what “changed our politics” is not “the Watergate office-apartment-hotel complex in Washington, DC” or its metonymic reference to the arrest of the five burglars at the Democratic National Headquarters on June 17, 1972, but a series of subsequent events that ultimately led to a likely impeachment and the actual resignation of a President for the first time in US history.

Dirven (1985: 96–97) sees metonymy as the metaphorical process whereby one thing comes to stand in place of something else due to their various intrinsic relationships: (i) a place and the people who live there; (ii) a symbol and the person it stands for; (iii) container and contained; (iv) the name of a producer or place and the product; (v) an article of dress and a person who wears it; and (vi) (the name of) an author and his work. And he observes that only the first four have become productive in the extension of the lexicon, and the metaphorical process is not operative for all metonymical relationships, because there is no constant relationship in the experiential world in need of a reflection in the lexicon. Peirsman and Geeraerts (2006: 276–277) have listed 23 metonymic patterns, which they distinguish as follows:

1. Spatial part and whole

2. Temporal part and whole

3. Location and located

4. Antecedent and consequent

5. Subevent and complex event

6. Characteristic and entity

7. Producer and product

8. Controller and controlled

9. Container and contained

10. Material and object

11. Cause and effect

12. Location and product

13. Possessor and possessed

14. Action and participant

15. Participant and participant

16. Piece of clothing and person

17. Piece of clothing and body part

18. Single entity and collection

19. Time and entity

20. Object and quantity

21. Central factor and institution

22. Potential and actual

23. Hyponym and hypernym

They admit that this is not “a complete and definitive list of metonymical types,” but that they merely “define an empirical basis for the analytical exercise.” In fact, their metonymic patterns have empirically opened a Pandora’s box, which explicitizes a multitude of logical relations between the elements of discourse or thought in terms of reasonableness or intelligibility. Apart from part and whole, part and part, simple and complex, space and time, cause and effect relations, there are temporal, spatial, physical, corporal, eventual, semantic, numerical, etc. relations, along with those of attribute and bearer, possessed and possessor, contained and container, clothing and wearer, material and product, product and trademark, located and location, quantity and thing, product and producer, object and function, etc. Almost any two linguistic or actual entities that are logically and conceptually related in any way can be used metonymically (substitutively) or symbolically to highlight the relationships and create new meanings.

Semiotically, according to Jakobson (1960: 369), linguistic elements substitutable in the same place in a syntagm are simply “equivalents,” not synonyms, since “equivalence” also includes contrast and other types of paradigmatic relationships. Thus, our attention has to extend further to how the substituted references collocatively influence other elements on the syntactic level and what messages and implications are being conveyed by the whole utterance, rather than simply identifying what entity is used to refer to another on the lexical level. The inflation example can be seen as a syntactic metonymy, since it rhetorically transfers an expression of a proposition into another that is conceptually similar and semantically synonymous, maintaining the logical relation (e.g. cause and effect, etc.) between a fall in value and a total monetary loss. The “incongruence” between the otherwise normal expression and our conceptualization of processes is what Halliday (1985: 322) calls “grammatical metaphor,” a rhetorical transference which is also realized by metonymy and synecdoche. The ham sandwich example contains a typical lexical metonymy of not “imputing human qualities to” things (Lakoff and Johnson 1980: 35), but it is also actualized by a syntactic metonymy, as a rhetorical transference of an expression in a proposition into another that is contiguously related on the association of the logical subject and the logical object of sandwich-ordering. Functionally, it is not simply a metonymy of using the logical object in place of the logical subject, but is a metonymy of using the grammatical object to substitute the whole attributive clause (“… who ordered a sandwich”), which, in turn, substitutes the head noun “the customer” on a higher syntactic level.

Kress and Hodge (1979: 7) emphasize that “the grammar of a language is its theory of reality,” since language consists of a related set of categories and processes, and the fundamental categories describe the interrelation of objects and events. By conceptualizing the customer in terms of the order he has made, the waiter focuses on thing rather than on person. While the waiter is not imputing human qualities to a thing, he is consciously or subconsciously dehumanizing the customer, who will most probably feel offended or even insulted if he hears the waiter referring to him as “the ham sandwich” instead of the usual “patron” or “customer” or simply “the man,” because, as Wardhaugh and Fuller (2015: 44) point out, some terms for referring can be “potentially offensive.” According to Lyons (1995: 298), two kinds of presupposition (existential and categorical) work at the same time: whoever uses the expression “the woman” or “the man” in an ordinary context presupposes the existence of a person on the one hand and categorizes the person into a particular sort on the other.

As a social semiotic, language evokes realities beyond the literal linguistic content and is a set of symbolic resources. Halliday (1985: 322) explains that part of knowing a language is to know the most typical “unmarked” way of saying a thing, but there are other possibilities “where the unmarked mode has been departed from and the speaker or writer has chosen to encode things differently.” In departing from the norm, the speaker’s choice of words for both the action-doers and the action processes is stylistically weighed and culturally and ideologically impregnated. Booth (1961, 1983 has consistently shown that even the most impersonal and impartial narration actually uses language to carry personal evaluations. Thus, by referring to things and people in deviant ways, the speakers’ codification becomes semiotically significant: we find more connotations of economic decline, ruthless depreciation, and consequent deprivation of one’s savings, and of the waiter’s conceptualization of the customer as a thing along with a possible implication that the waiter finds the customer is anxious or impatient waiting.

Metonymy is also a cognitive and discursive process, and is essential for understanding the way in which “creative meanings arise” (Hidalgo-Downing and Mujic 2013: 162). It is true that creative meanings arise right when we analyze the cognitive paths along which one term is used for another, a logical “acted” is used for a logical “actor,” a syntactic element is used for a syntactic structure, etc. Hence, metonymy is not only something that “we live by,” but also something that metaphors live by, since it is probably “even more basic” to language and cognition, and “a special case of what Langacker (1987: 385–386) calls activation” (Barcelona 2000a: 4). Langacker (1993: 30) believes that metonymy is prevalent because “our reference-point ability is fundamental and ubiquitous,” and it precedes metaphor because of a “useful cognitive and communicative function.” The metonymic mapping causes the mental activation of the target domain (Kövecses and Radden 1998: 39), and metonymy can identify the target of a metaphor, limit the correspondences between the domains, or expand and create new meanings (Uriós-Aparisi 2009: 110).

In semiotic terms, Wales (2014: 268) sees metonymy as an “indexical sign,” since there is “a directly or logically contiguous” relationship between the substituted word and its referent: a product like floor-cleaner in TV advertising is indexical of the domestic lifestyle also depicted. Jakobson (1956) proposes that a discourse may develop along two different semantic lines: one topic may lead to another either through their similarity or through their contiguity, and they are the “most condensed expression” (1956: 76). In relating metonymy to the linguistic practice of syntagmatic combination and to the literary practice of realism, he points out that the importance of metonymy to the so-called “realistic” trend has not been fully realized: along the path of contiguous relationships, the realistic author metonymically digresses from the plot to the atmosphere and from the characters to the setting in space and time, such as Tolstoy’s attention focused on Anna Karenina’s handbag at the time of her suicide (1956: 78).

Although Lakoff and Johnson (1980) make no mention of “semiotics” in the study of either metaphor or metonymy, their observation of “symbolic metonymies” already paved the way for semiotic studies: symbolic metonymies that are grounded in our physical experience can “provide an essential means of comprehending religious and cultural concepts” (Lakoff and Johnson 1980: 40). Al-Sharafi (2004) proposes a semiotic approach and argues that metonymy is “primarily the dimension of context” on a further level of signification of things and objects. On his belief that a semiotic tridimensional model of metonymy is related to another tridimensional model of text that realizes text as language, text as cognition and text as context, he provides a definition of metonymy: “a process of representation in which one word or concept or object stands for another by contiguity or causality” (Al-Sharafi 2004: 1).

The function of metonymy has been employed in narrative construction. Chun and Zubin (1995) have found that a systematic body part metonymy is frequent in the experiential construction, since the body part is the relevant organ to which the stimulus must come in order for it to be experienced (1995: 322). A few other studies have focused on the role of metonymy in thought and in discourse (Forceville and Uriós-Aparisi 2009; Sobrino 2017). Yet, all studies of metonymy focus on its being referential shorthand or hidden communicative shortcuts.

3 Metonymy as cognitive roundabout

Metonymy has served a wide range of cognitive functions in both everyday talk and literary texts. Unlike metaphors, which have already been identified in different discourse contexts and held to be “sites of intensive work relating to the central discourse purpose” (Cameron and Stelma 2004: 107), metonymy has not attracted equal attention on the discourse level.

Papafragou (1996) observes that metonymy is widely defined as “referential shorthand” whose success is grounded on empirical associations among objects and which acquires implications of closeness and belonging, especially when used within professional and other in-groups. In his book entitled Metonymy: Hidden shortcuts in language, thought and communication, Littlemore (2015) states that metonymy is a kind of “communicative shorthand” that allows people to “use their shared knowledge of the world to communicate with fewer words than they would otherwise need” (2015: 5) or to use a relatively simple or concrete entity to “provide easy access to an entity that is much more complex or abstract” (2015: 65).

Contrary to these observations, my study has found that metonymy in literary discourse may go the other way round, employing multi-level or chain metonymies that are never cognitively “short.” Instead of communicating with fewer words, literary authors use more or far more words than they would otherwise need; instead of using relatively simple or concrete entities to provide easy access, they use much more complex or abstract entities to provide roundabout and difficult access in their fictional representation.

At the outset, I have to admit that linguistic analysis of literary texts is not without problems. First and foremost, as Crystal (1972: 112) points out, much linguistic analysis of literature has not been received well because “linguists take texts which seem interesting and problematic to them” and “they often forget that the text, or the problems, may not be of comparable interest to the critic.” While it is not possible to thoroughly appreciate the literary critic’s problems and position, it is possible to address issues common to literary criticism and cognitive stylistics/poetics, such as “defamiliarization” (Shklovsky 1965 [1917]) and “refamiliarization” (Stockwell 2002: 79, 2019: 107). I am not choosing a text that seems particularly problematic or interesting only to some people, or only adequate for my own purposes, but a very popular one that may be of equal interest to almost all: one of O. Henry’s best-known short stories, “The Cop and the Anthem.”

4 Metonymic devices in fictional narrative

The plot of the short story is not complicated, but it involves a number of semiotic techniques to represent the protagonist’s psychological experiences from the beginning to the end. The whole process from the time the protagonist, Soapy, makes the decision to get assured food and shelter during the winter and develops a series of offensive tactics to get into trouble and arrested until the point when he has made up his mind to turn over a new leaf but is categorized as a potential criminal and “caught” unawares, O. Henry uses various types of metonymy to represent the setting, actants, and processes of actions.

Neither the protagonist nor the cop, nor indeed the anthem, is the theme of this short story. The title “The Cop and the Anthem” is a metonymy that leads the reader to expect anything but a tramp or an attempting offender. In fact, this discrepancy might be what O. Henry intentionally meant for his readers in 1904. Discursively, it falls into the type of part for whole metonymy, since the cop and the anthem are only peripheral elements in the story. Just as the referential “shorthand” of ham sandwich can conveniently pick out a customer among many, the title of this short story paradigmatically makes a difference, as all others do, on the contents pages of The four million (Henry 1906) and The complete works of O. Henry (Henry 1953).

5 Conclusions

Metonymic devices in this short story are not blatant violations of semantic constraints that require some repair mechanism for proper interpretation. Rather, they are manifestations of figurative language that depends on semantic representation structures consistent with the underlying knowledge representation (Markert and Hahn 2002). Nor are they simply substitutions of terms or matching of counterparts that conceptually foreground the sources and background the targets (Barcelona 2000b) or highlight one (sub)domain over another (Uriós-Aparisi 2009) on the lexical and syntactic levels. They are systematic textual exploitations of metonymy as a semiotic resource that offers far more functional equivalents that broaden our vision. They are not prototypical metonymies in which the target becomes conceptually prominent and the source is backgrounded, or conventional metonymies that build on obvious logical relations and cognitive associations. They are, however, semiotic means of explicitizing complex sociocultural associations between the source and the target on diverse and inconstant metonymical relationships that are not normally obvious in the experiential world. They have highlighted the changes in salience or viewpoint for their communicative function and functional motivations and made the text more informative and symbolically richer. By providing stylistic deviance and creating artistic estrangement, O. Henry presents a defamiliarized fictional world in which almost everything requires more cognitive efforts from the reader, who is artfully led into re-conceptualizing the actual world through fictional representation. In perceiving reality in roundabout ways, the reader becomes refamiliarized with a world that is more meaningful than otherwise. Thus, contrary to the primary function of providing “referential shorthand” or “communicative shortcuts,” the primary function of metonymy in literary narrative is to provide “cognitive roundabouts,” which exploit semiotic resourcefulness in literary creation and offers more difficult access to the text only for artistic purposes.