Metonymy, Metaphor, and Meaning in Inflected Phrasal Verbs: A Corpus Analysis


Though cognitive linguistics has greatly expanded the understanding of how conceptual metaphors and metonymies can structure thought, research on the linguistic environments wherein these figurative words appear is limited. The interactions between these two phenomena (Goossens, 1990) raise more questions about the interaction between their cognitive and grammatical features as well. Using modern corpora, this grammatical and abstract interplay can be studied by looking at the two together. The Corpus of Contemporary American English (Davies, 2008) shows that the four main inflections (the root, -s, -ed, and -ing) of phrasal verbs involving “bottle” and “hand” often correlate with different conversational or semantic environments, as well as corresponding with an orientational conceptual metaphor specific to their particle. Examining the semantic environments of necessarily metonymic or metaphorical constructions provides a comparative baseline for broader research into both the production and conceptualization of metaphor and metonymy.

Table of Contents: 


    In the past, figurative expressions such as metaphors were viewed by both linguists and the public as merely flowery, literary language. A Grammar of Metaphor (Brooke-Rose, 1958) marks the first time an in-depth study was done on the grammar of metaphors, which still focused on their appearances in poetry and literature. However, since the 1980 publication of George Lakoff and Mark Johnson’s Metaphors We Live By, the processing and production of figurative language has received more attention. This work laid the foundation for the cognitive linguistic theory of conceptual metaphor; even today, the realized grammar or syntax of the phenomena is not studied as widely. However, with modern corpora, it is possible to study the actual linguistic environments of metaphors and metonymies. The researcher is able to see which words pair with or appear near each other by using a collocation search and noting the frequency of parts of speech carrying figurative meaning, both of which help to reveal the linguistic environments of figurative elements. Having concrete, grammatical knowledge of the appearance and use of metaphor and metonymy has implications for second language pedagogy as well as providing a mutualistic link between cognitive and theoretical linguistics.

    Literature Review

    Definitions of Metaphor and Metonymy

    Metaphor and metonymy are often studied together, and while both are somewhat hidden processes – commonly understood or idiomatic examples are often not recognized as figurative expressions at all by the speakers using them – metonymy gets much less attention as an independent phenomenon while also being less well known to the public as a whole. The two processes, however, are quite distinct, though their correspondence with each other is often noted, making their distinction fuzzier when they work on a cline (or gradation) from one to the other. In “Metaphtonymy,” his paper demonstrating the interaction between the two, Goossens (1990) explains: “The crucial difference between metonymy . . . and metaphor is that in a metaphorical mapping two discrete domains are involved, whereas in a metonymy the mapping occurs within a single domain” (p. 325).

    The essential difference between the two is that a metaphor takes two completely different things (a source and target) and applies the features of the more concrete one to the abstract one. Metonymic figuration, however, draws attention or changes the focus of one thing by honing in on the features of something very closely related to it, keeping the connection intra-domain. An example of a metonymic utterance would be using “wheels” to stand in for “car,” as in the utterance “I’ve got a new set of wheels” (Lakoff & Johnson, 1980, p. 36). The replacement of wheels for the car as a whole effectively puts the focus on the speed or attractiveness of the car, but does not abandon the vehicle as whole – making the vehicle both the source and target domain.

    Still, in phrases and syntactic structures larger than the single words that carry figurative meaning, the two often meld (Deignan, 2005; Goossens, 1990). The four main patterns identified by Goossens are metaphor from metonymy, metonymy within metaphor, metaphor within metonymy, and demetonymization within a metaphor, with the first two being the most common (1990, p. 337). A metaphor can be lifted from an original metonymy, seen in Goossens data in idioms such as “bite one’s tongue off,” and “catch someone’s ear” (p. 333-334).

    The Conceptual Metaphor Theory

    The conceptual metaphor theory uses the basic definition of metaphors above to identify the maps of common, figurative concepts that the mind constructs, which are used by individuals and cultures at large to process ideas. As established in the literature, the names of conceptual metaphors are written in small caps for clarity that a given figurative expression is canonized as a mapping. Two common conceptual metaphors are ARGUMENT IS WAR and HAPPY IS UP; these appear in language in utterances such as, “His criticisms were right on target,” and “My spirits rose” (Lakoff & Johnson, 1980, p. 4, 15). Their seminal work demonstrates that the cognitive process is revealed when people attempt to speak about the abstract – or target – concept, and the typical usage is the metaphorical one. That it is the typical usage demonstrates how

    we don’t just talk about arguments in terms of war. . . we gain and lose ground. We plan and use strategies. . . . Many of the things we do in arguing are partially structured by the concept of war. (Lakoff & Johnson, 1980, p. 4)

    While ARGUMENT IS WAR is a structural type of metaphor, orientational metaphors such as HAPPY IS UP and HAVING CONTROL, OR FORCE IS UP (“I am on top of the situation,” “I have control over her,” (Lakoff & Johnson, 1980, p. 15), follow the same core premise: “the concept is metaphorically structured, the activity is metaphorically structured, and, consequently, the language is metaphorically structured” (Lakoff & Johnson, 1980, p. 5).

    Independent conceptual metonymies, however, have also been identified and recently studied by Radden and Kovecses (1999). Their study features a discussion of the conceptual metonymy instrument for action – seen in words such as the infinitive verbs “to ski” and “to hammer” – in a section devoted to the mental processes that create this kind of language (p. 37). In The Interaction of Metaphor and Metonymy in Noun-To-Verb Conversion, a paper on this phenomenon, Kuczok (n.d.) notes: “The phenomenon of conversion or zero derivation is a word formation process in which a given word changes its category without changing its form. The cognitive processes behind this linguistic phenomenon are often ascribed to metonymies” (Abstract).

    Processes such as these offer further clues to the cognitive nature of figurative utterances in language. It may also especially aid in the logic underlying accepted conceptual metonymies, as the INSTRUMENT FOR ACTION concept named above is actually reversible, converting into another conceptual metonymy, ACTION FOR INSTRUMENT:

    Participants are converted into verbs and predicates are nominalized. Noun-verb conversion and nominalization can therefore be seen as two complementary morphological processes leading to the two types of reversible metonymies. What makes these morphological derivations special types of metonymy, however, is their conflation of vehicle and target: the original word class describes the metonymic vehicle and the morphologically recategorized form expresses the target. (Radden & Kovecses, 1999, p. 37)

    “Screwdriver” and “pencil sharpener” (p. 37) were specifically identified by the authors as examples of words formed through this process, showing the complex interplay between both verbalization and nominalization.

    Corpus Study into Uses of Metaphor and Metonymy in Language

    Corpora work has been done on metonymy and metaphor independently, as well as on the gradients between the two, with a special interest in identifying grammatical patterns or fixedness. In a study of 1000 citations of metaphors from the human body, restricted to nouns, Deignan (2005) identified 3 categories of fixedness: free standing, relatively fixed, and those that were always qualified. The free range describes utterances such as “everyone lent a hand,” (p. 160) while an example of the second, relatively fixed set, is the metaphorical use of “shoulder” to mean “have responsibility” – this usage is heavily restricted to the plural, “shoulders,” and features strong grammatical fixedness with the necessity of “his/her, etc.” preceding it (p. 160).  The third group describes bodily metaphors that always appear in a fixed form, such as “heads of state” (p. 161). The ability of modern mega corpora to display frequency statistics as well as be filtered heavily for certain linguistic parameters make it an invaluable tool for uncovering deep patterns in language use.


    To dig deeper into the figurative construction of language, particle phrasal verbs created by the noun-to-verb conversion were studied for their potential insight into how a primarily metonymic process may take on metaphorical elements when parts of speech join together. The resulting semantic environments that featured these constructions were studied for differing grammar, to see if certain semantic domains adhere to certain inflections of the metonymic verb, or appear in higher frequency. All searches were run in the Corpus of Contemporary American English, a 450 million word collection of texts and recorded speech from a variety of sources, such as television, fiction, and newspapers. It was mined for instances of the phrasal verbs “to bottle up,” “to hand in,” “to hand down,” and “to hand back.” The two root words, chosen for their separateness in both source domain (human body versus inanimate container) and use of particles, were studied for correspondences between their semantic environments, verbal inflections, and their degrees of metaphoricity. Two related phrasal verbs, “to hand out” and to “to hand over,” were excluded from this particular study in order to be used comparatively in a future study using more than two source domains. Each main verb was searched for in each inflection (unmarked, -s, -ed-, and -ing) using a collocation search to find instances of the particle (up, in, down, and back) appearing within four places after the main verb. Up to 50 results were used as examples for each phrasal verb; however, many searches returned fewer than the maximum 50 results. Percentages were taken to show the overall frequency of the collocations within the overall verb results, and the 50 +/- of actual phrasal verb examples were marked for the conversational topic or semantic area in the samples in which they were found.


    To compare frequency percentages of the phrasal verb forms, the overall frequency of the root verb forms were noted first, and listed in the top column of all tables for comparison. The proportion of the inflected forms, across the differing source domains of the human body and an inanimate container, are not parallel. The simple past appears the most in both source domains, yet “handed” appears most across all 8 inflections studied, with 12,737 occurrences versus 401 occurrences of “bottled.” Though the combined appearance of all forms of “bottle” nowhere near match the frequency of those of “hand” – an imbalance noted in the methodology section – no other forms came close to the prevalence of “handed”; the next most common inflected form is the present progressive “handing” with only 3,563 occurrences. These numbers also create a reference to account for possible greater percentages of semantic environments in the more frequently occurring inflections, such as the simple past mentioned above. The number of appearances of the phrasal constructions as a whole was also compared to their main verb’s overall appearance to see if the use of the naturally metonymic verb was most commonly employed as a phrasal verb. The inflections, particles, and semantic environments of each phrasal verb are now discussed independently.

    To Bottle up

    The first search was for the different inflected forms of the base phrasal verb, “to bottle up.” See Table 1. A collocation search was done for the word “up” following each verb form within four words. After uninflected “bottle,” the particle appears 24% of the time, after “bottles” 19% of the time, after “bottled” 52% of the time, and after “bottling” 36% of the time. These semantic environments account for a greater proportion of their overall main verb results than the various hand phrasal verbs, most likely because the verb form of “bottle” takes fewer particles altogether and is very likely to appear as a metonymic phrasal verb over other structures.

    Across each inflected form of the phrasal verb, only 9 main conversational topics were identified in the collected data, all within three larger categories, which are abstract containment, physical containment, and using containment to thwart. Of the subtopics, 41% contain emotions and 14% involve the government, with these domains taking up 55% of the examples. These topics are not divided equally throughout the inflections, however; both are largely carried by “bottle” and “bottled,” and while both carry 21 instances of contained emotions, “bottled” is 3 times more likely to be found in a discussion about the government than the unmarked “bottle.” These examples give credence to the theory that semantic contexts use certain inflected forms more than others. Another interesting finding for this phenomenon, which may offer additional insight into the conversational abstraction often accompanying the phrasal verb, is that only the unmarked “bottle” includes instances of an actual liquid or food being placed and sealed in a bottle, accounting for 16% of results.

    To Hand Down

    Even as the source domain changes to the human body, and the meaning of the previous particle is inverted, the results for “to hand down” still show preference for certain conversational topics. See Table 2. Just two overarching categories account for 75% of the discourse topics in the examples. The first is named in the table as family sharing and encompasses traditions and items being given to younger people, and the second is listed as verdict and order, including concepts such as court decrees, governmental laws, and prison sentences. However, this strong prevalence does not distribute itself evenly across all inflections of the verb. As seen in the much lower frequency of “bottles” across both semantic topics and overall use, “hands” manages not to have any examples involving family sharing; half of its very few results fall under the verdict and order class, accounting for 17% of instances.

    Only two other major categorical domains were found for this phrasal verb; the figure lists them as history and physical giving, which account for 14% and 9%, respectively. Though all the categories have the ability to include a physical object and are not restricted to abstract notions, only physical giving is restricted completely to a literal act of moving control or ownership of something from one person to another, and it is used much less often, as well as preferring the unmarked verb inflection, using it 57% of the time. “Handed” has one instance of a physical object while “handing” has none – the results of these two share other similarities throughout this data set as well. Both appear much more often in the corpus as a verb, collocate more frequently with “down,” and 50 examples of the definite phrasal verb construction were sampled from the collocation. Even their semantic environment frequencies almost exactly match, differing only by 2 results at the highest.

    The distinctness of the four categories, however, is still logical when one compares them metaphorically. Through the medium of the particle “down,” all but the literal, physical giving category are united by conceptual orientational metaphors. Just as Lakoff and Johnson (1980) identified HAVING CONTROL OR FORCE IS UP; BEING SUBJECT TO CONTROL OR FORCE IS DOWN, the semantic environments of the results adhere to this and other principles by way of the particle and not the root verb. Verdict, family sharing, and the less common history branch of the results all involve a higher authority bestowing some kind of object, idea, or decision onto a lower entity – benevolent or not. The verticality of such a transfer seems embedded in the particle; the metonymic verb driving the construction seems to account for the specific types of transfer, a human to human bequeathing, a metonymic personified entity (i.e. “the court”) making decrees to “the people,” as well as a more abstract human knowledge “posterity” sharing.

    To Hand In

    This phrasal verb marks a return to a small number of overarching categories containing specific and recurring semantic environments, similar to the pattern seen in “to bottle up” but not in “to hand down.” With 10 specific areas, 8 fall under the umbrella of only 2 larger categories; the umbrella terms named in the table as piece of writing and surrender object encompass a full 83% of the particle phrasal verbs results, and account for 15% of all simple collocations for the preposition “in” and the verb “hand.” See Table 3. This dataset is also not distributed equally however; piece of writing accounts for 70% of its occurrences here. This heading is subdivided into school/homework, business, and the most specific category thus far, letters of resignation. Letters of resignation account for 17% of all these particle phrasal verb results alone. An interesting note about this very particular object of conversational discourse is its strong relation to the second highest umbrella category, the surrender object branch, as in its very nature a resignation letter is the surrendering of an object as well, raising interesting questions about how the physical object of the letter can be metonymic for the job, position, etc., that is being surrendered.

    Another interesting result from this phrasal verb is that all inflections of the verb are heavily weighted toward the top section of the table, with only sparse occurrences making up the second portion containing the surrendered objects and miscellaneous distinctions of physical moving and political realia (ballots, voter identification, etc.) – the final two accounting for only 10% and 6%, respectively. As seen in the previous results, the –s inflection is stark in its low frequency of use and the small percentage overall of uses of this construction to describe unspecialized, literal movement. In the concept of “returning” the particle “back” inspires, no immediate orientational metaphors arise; however, a deeper structure may underlie this construction through the employment of idealized cognitive or similar models, which is outside the scope of this research.

    To Hand Back

    Similar to the preceding discussion, “to hand back” featured strong associations with paper and written correspondence. It, too, has a piece of writing heading under which falls 22% of its occurrences. See Table 4. However, this umbrella term is dwarfed by return physical, the next categorical section of the results. While the metonymic verb “hand” meaning “give to” necessarily involves transfer and therefore movement, many previous uses displayed here showed in phrasal verbs that it can often be subsumed by abstract concepts or metonymic objects of abstraction, such as the resignation letters of “to hand in.” The physical results of “to hand back” are often very strongly physical and obvious; in this section, miscellaneous physical objects being returned account for 67% of results, weapons account for 15%, money accounts for 9%, and drugs or alcohol accounts for 8%. The nondiscriminatory uses of this verb when it comes to physical objects is unsurprising given its high profile in the corpus collocation searches, coming second in terms of overall results, after only “hand down,” and being less restricted due to underlying orientational metaphors than its cousin construction. That it has 3 heavily distinct physical objects following the hodgepodge of the first is interesting, especially considering the possible “less savory” characters money, drugs/alcohol, and weapons may take in discourse. This result also marks a change in direction for the –s inflection, which in this construction occurs frequently in both the general objects category of return physical, and is tied for most occurrences in the drug/alcohol category, along with “handed.” This conversational topic has no hits in unmarked “hand.”

    The least common and most starkly discriminating topics in this data set fall under the heading of control. In this section, the semantic environments of land, person, and miscellaneous other were noted – with almost all hits occurring in the unmarked inflection “hand,” with “handing” as a second. Out of these results, control of person was the most common, accounting for 42% of the umbrella category as a whole.  “Handed” was very strongly unaffiliated with this concept; however, it accounts for the most overall verb results by far and in this set is the most frequently collocating with “back,” as it carries only 11% of occurrences.

    Hearkening back to the metaphorical results of “to hand down,” one hypothesis for the likelihood of the particular discourse topics “to hand back” is a horizontal orientation between the participants in the discourse or the persons being discussed. The returning nature embedded in “back” necessitates a prior, if unmentioned, “first giving,” which opens the door to many other related expressions, even featuring different parts of speech such as “handing something back and forth.”


    As seen in previous studies, metaphors and metonymies do often emerge in distinct and semantically predictable environments. All the phrasal verbs strongly correlated to specific conversational topics, with many of the identified conversational topics showing slight to strong preference for certain inflections of the verb – the –s marked inflection both the least common and most discriminating inflection. Some particles, specifically “down,” also appeared to correspond with recognized orientational conceptual metaphors, many of which then also showed preference for certain verb inflections in the discourse examples in which they were found.


    • Brooke-Rose, C. (1958). A grammar of metaphor. London: Secker & Warburg.
    • Davies, M. (2008-) The Corpus of Contemporary American English: 450 million words, 1990-present. Retrieved from
    • Deignan, A. (2005). Metaphor and corpus linguistics. Amsterdam: John Benjamins Publishing.    
    • Goossens, L. (1990). Metaphtonymy: the interaction of metaphor and metonymy in expressions for linguistic action. Cognitive Linguistics, 1(3), 323-342. doi:10.1515/cogl.1990.1.3.323
    • Kuczok, M. (n.d.). The interaction of metaphor and metonymy in noun-to-verb conversion [Abstract]. In B. Bierwiaczonek, B. Cetnarowska, & A. Turula (Eds.), Syntax in cognitive grammar (41-54). Częstochowa: Wyższa Szkoła Lingwistyczna.
    • Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of Chicago Press.
    • Radden, G., & Kovecses, Z. (1999). Toward a theory of metonymy. In K. Panther, & G. Radden (Eds.), Metonymy in language and thought (17-59). Amsterdam: John Benjamins Publishing.

    Table 1: Frequency of use of phrasal verb “to bottle up” in 4-place collocation search of “up” and “bottle,” and their semantic contexts

      BOTTLE (301) BOTTLES (16) BOTTLED (411) BOTTLING (26)
    Contain abstract        
    Emotions 21 1 21 3
    Essence, “you” 8   1  
    Contain physical        
    Liquid 8      
    Traffic     1 2
    Restrict access     4 1
    Block/tense body 2   2  
    Contain to thwart        
    War 2   4  
    Sports 6   3 2
    Government 3 1 10 2
      50/73 ‘up’ -/+ 4 2/3 ‘up’ -/+ 4 50/213 ‘up’ -/+ 4 10/12 ‘up’ -/+ 4

    Table 2: Frequency of use of phrasal verb “to hand down” in 4-place collocation search of “down” and “hand,” and their semantic contexts

      HAND (1831) HANDS (2627) HANDED (12737) HANDING (3563)
    Family sharing 12   17 18
    Verdict, order 13 6 21 23
    History 1   11 9
    Physical giving 8 5 1  
      34/52 ‘down’-/+ 4 11/33 ‘down’-/+ 4 50/944 ‘down’-/+ 4 50/71 ‘down’-/+ 4

    Table 3: Frequency of use of phrasal verb “to hand in” in 4-place collocation search of “in” and “hand,” and their semantic contexts

      HAND (1831) HANDS (2627) HANDED (12737) HANDING (3563)
    Piece of writing        
    School/homework 6 3 20 11
    Texts, business 8   8 5
    Resignation letter 4   12 3
    Surrender object        
    Self   1    
    Money   1    
    Personal ID/badge 1 1    
    Weapons 1   1 4
    Keys (take leave)   1 2 2
    Political realia 3   5 4
    Physically move 3 2 2  
      26/56 ‘in’ -/+ 4 9/29 ‘in’ -/+ 4 50/458 ‘in’ -/+ 4 29/105 ‘in’ -/+ 4

    Table 4: Frequency of use of phrasal verb “to hand back” in 4-place collocation search of “back” and “hand,” and their semantic contexts

      HAND (1831) HANDS (2627) HANDED (12737) HANDING (3563)
    Piece of writing        
    School/homework 1 1 4 4
    Texts, business 10 10 7 7
    Return physical        
    Gen. objects 18 27 21 21
    Money 3 3 1 5
    Weapons 3 5 10 2
    Drugs, Alcohol   4 4 3
    Over land 5   2 4
    Of person 7   1 3
    Other 3     1
      50/52 ‘back’-/+ 4 50/118 ‘back’-/+ 4 50/543 ‘back’-/+ 4 50/130 ‘back’-/+ 4