Introduction

2The aim of the present study is to compare denominal verbs whose base is either a PN or a common noun (“CN” from now on) and see whether the analyses of their behavior in derivation can shed light on the difference(s) (and similarities) between the two categories: does the verbal derivation of PNs involve different processes from the verbal derivation of CNs? In other words, the present study explores the significance of the PN/CN variable on verbal derivation.

3The study relies on the decomposition of verbal derivation into relevant steps so as to reverse-engineer the path which leads from the nominal base to the meaning of the verb. Examples (1) and (2) illustrate the type of analysis undertaken for the study:

• 3 Oxford English Dictionary online, see Section 2.

(1a) Verb – Bogart: “to force, coerce; to bully, intimidate”. [OED3 2021]

(1b) Morphology – Formed by conversion, from the PN Bogart.

(1c) Base – Humphrey Bogart (1899–1957), U.S. film actor who sometimes played “tough characters”. [OED 2021]

• 4 I consider here that the derivation process involves both agent for event and event for characteris (...)

(1d) Metonymic patterns – agent for event + event for characteristic + characteristic for action⁴.

In example (1), the event <play the role of bullies> is selected from the life of Humphrey Bogart. The characteristic <bully> is in turn selected within the event to become, through the process of conversion, a verb denoting the action involving this characteristic (see Section 3.2. for further details about metonymic patterns and their selection).

• 5 The full definition of the OED is as follows: “to cook professionally; to work as a chef; to presid (...)

(2a) Verb – chef: “to cook professionally”. [OED 2022]5

(2b) Morphology – Formed by conversion of the CN chef.

(2c) Base – chef: “a professional cook, esp. the head cook in a restaurant”. [OED 2022]

(2d) Metonymic patterns – agent for action.

Example (2) is more straightforward as the verbal derivation only includes one metonymic pattern in which the main semantic feature of the CN is converted into an action.

1. Theoretical background

5In order to compare PN verbal derivation to CN verbal derivation, I first describe how PNs are understood in this study, inherently and relatively to CNs (Section 1.1.). I then present the morphological processes at work in the data under scrutiny (Section 1.2.1.) before moving on to the core approach, based on Janda [2011], which postulates that metonymy accompanies word-formation processes (Section 1.2.2.).

1.1. What is a proper name?

6There are many linguistic theories on PNs, some of which take their roots in other disciplines, such as philosophy. The purpose of this article is not to present a description of the many theories about PNs but to offer an overview of their definition. Some major theoretical trends as well as terminological choices are presented in this first section.

9Following Štekauer [1997], Gary-Prieur [2016] and, more recently, Philippe [2020], I consider that both CNs and PNs have a semantic content, which derives from both encyclopedic knowledge and semantic features. As a result, the existence of semantic features or the reference to encyclopedic knowledge are not defining criteria to discriminate between CNs and PNs. Philippe [2020: 449] interestingly concludes that PNs and CNs can be distinguished based on their opposed categorial dynamism:

• 15 “Proper nominals, which are modelled after common nouns (morpho-syntactically but also semantically (...)

Le nominal propre, dont le nom commun est le modèle (morphosyntaxique mais aussi sémantique), va progressivement se doter d’un potentiel de catégorisation, mais toujours en sens inverse du nom commun : issu de la connaissance d’un référent extralinguistique, c’est par le référent initial du [nom propre] que se construit progressivement une classe, une catégorie discursive, et parfois une catégorie en langue associée au nom propre.15

11To sum up, the present study tends towards the meaningfulness thesis on PNs but avoids the controversy around the issue of meaning by simply considering that both CNs and PNs have semantic content. Their differences lie in their dynamics. As a process, verbal derivation is by nature dynamic. As a result, the comparison between PNs and CNs in verbal derivation should allow to test Philippe’s conclusion above.

1.2. Perspective(s) on denominal verbal derivation

12Before delving into two elements of verbal derivation, namely morphology and metonymy, it is worth noting that the present study fits into a cognitive linguistic framework and adopts an onomasiological perspective. Štekauer [2005: 207] opposes the onomasiological approach of word-formation to the semasiological approach:

The semasiological […] method, proceeding from form to meaning/concept, concentrates on the analysis of the already existing word-stock. The onomasiological […] method, which takes the opposite direction and studies the naming act, has long been relegated to the periphery of research in works on English word-formation.

Thus, the intended meaning is considered the base for derivation. As a result, the speaker is viewed as “a linguistic innovator” and “the naming act is a cognitive phenomenon relying on the intellectual capacities of a coiner” (Štekauer [2005: 212-213]).

13The aim of the study is to compare verbs according to their base and explore the following hypothesis: the categorial distinction between PN and CN is significant when considering the cognitive processes at stake in denominal verbal derivation in English. As a result, and following Janda [2011], two variables need to be specified here: verbal morphology and metonymic patterns. I briefly present the two in the following sections and give more details on their encoding in Section 3.

1.2.1. The morphology of denominal verbs

• 21 Prime minister: “To govern. Also intransitive: to act as prime minister” [OED 2022].
• 22 In the case of poor / the poor, the conversion is only partial as the syntax and semantics of the s (...)

(3a) from N to V: a prime minister >> to prime minister21
(3b) from V to N: to call >> a call
(3c) from A to V: better >> to better
(3d) from A to N: poor >> the poor22

16Suffixation is a type of derivation that consists in adding a bound morpheme to the right of a stem. According to Plag [2003: 116-117], there are four verb-forming suffixes in English, illustrated in the examples below (from VdenomEN):

• 23 Funambulate: “To walk or perform on a rope stretched between two points at some height above the gr (...)
• 24 Aladdinize: “To transport or transform as if by magic” [OED 2021].
• 25 Princify: “To make into a prince; to make princely” [OED 2019].
• 26 Piecen: “To join or piece together; to repair; spec. to rejoin (broken threads or ends) in spinning (...)

(4a) funambule >> funambulate23
(4b) Aladdin >> Aladdinize24
(4c) prince >> princify25
(4d) piece >> piecen26

All these suffixes do not show the same productivity, as Plag [1999: 117-118] underlines:

In comparison to -ize, the suffixes -ify and -ate seem to be more severely restricted, in that -ate has a high global productivity, but the probability of encountering new forms is rather low. Conversely, -ify has a higher chance of occurring in new formations, but the sheer number of these instances is rather low.

18Prefixation, as opposed to suffixation, is a type of derivation that consists in adding a bound morpheme to the left of a stem. In English, prefixation rarely modifies the class of the base. Bauer [1983: 217] lists a few class-changing prefixes most of which are verbalization prefixes as illustrated in (5) (from VdenomEN). As a result, prefixation is relevant in denominal verbal derivation:

• 28 Betail: “To deprive of the tail” [OED 2019].
• 29 Enshade: “To envelop in shade” [OED 2018].
• 30 Embus: “To mount a bus or transport vehicle” [OED 2022].
• 31 Declutch: “To disengage the clutch of a motor vehicle” [OED 2018].
• 32 Disembargo: “To release from embargo” [OED 2022].
• 33 Unmute: “Music. To cause (a musical instrument) not to be muted. Also intransitive: (of a musical i (...)

(5a) tail >> betail28
(5b) shade > enshade29; bus >> embus30
(5c) clutch >> declutch31
(5d) embargo >> disembargo32
(5e) mute >> unmute33

Bauer [1983] also adds non- to this list (as in non-stick) although it is not present in my data.

19We should also add to Dixon [2008]’s list of verbalization processes backformation as defined by Bauer [2003: 39]: “cases where the element subtracted is (or looks like) a morph with an independent existence elsewhere in the language, and especially where the process is a derivational one”. Example (6) (from VdenomEN) illustrates cases in which the end of a noun, here -er, is reanalyzed within the -er suffixal paradigm and removed to form the verb.

• 34 Buttle: “To pour out (a drink)” [OED 2021].

(6) butler >> buttle34

Here, I consider backformation a type of derivation because I take the perspective of the “coiner” who takes a noun and turns it into a verb.

20Finally, one last process which is considered here is clipping as it can also involve one base-noun. Clipping refers to the process by which phonological or morphological material is removed from a lexeme. Bauer’s definition [1983: 233] is however problematic considering the data under scrutiny: “process whereby a lexeme (simplex or complex) is shortened, while still retaining the same meaning and still being a member of the same form class”. Before delving into the issue, I give in (7) some examples of clipping (from Bauer [1983: 233]):

(7a) bi << bisexual
(7b) jumbo << jumbo jet
(7c) ‘Fro << Afro
(7d) Cong << Viet Cong

21Bauer’s definition and examples entail class-stability, from N to N or V to V, making it hard to define clipping as a potential verb-forming process. However, the denominal data collected here includes the examples in (8):

• 35 Green’s Dictionary of Slang, see Section 2.

(8a) fizz: “to be an informer” << fizgig: “a police informer” [GDS35 2022]
(8b) scoob: “to eat, usu. to eat snacks” << Scooby Doo: “TV cartoon character […] and his Scooby Snacks” [GDS 2022]

22The data under scrutiny mostly includes cases of conversion and -ize-suffixation and marginally other types of suffixation, prefixation, backformation and clipping. My data analysis (Section 4) will focus on the main processes only with a particular attention paid to the metonymic patterns at work.

1.2.2. Metonymy: a cognitive approach to word-formation

23The study of metonymy takes its roots in the classical rhetorical tradition in which it was considered one of the major figures of speech (Panther & Thornburg [2007: 237]). It was then defined as a “stand for relation in which the name of one thing (henceforth, the source or vehicle) is used to refer to another thing (henceforth, the target) with which it is associated or to which it is contiguous” (Panther & Thornburg [2007: 237]). More recently, metonymy took a more significant part in linguistic research, especially within the framework of cognitive linguistics. In the present study, metonymy is understood as defined by Kövecses & Radden [1998: 39]:

Metonymy is a cognitive process in which one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same domain, or ICM [Idealized Cognitive Model].

25One convincing example of a study of metonymy as a word-formation process is proposed by Janda [2011]. It presents a view in which “the semantic relationship between base and derivative [is] ruled by metonymy” (Bauer [2018: 10]). Her study has raised some criticism (Brdar & Brdar-Szabo [2013, 2014]); Bauer [2018: 10] summarizes the debate as follows:

Either we accept that conversion is a matter of metonymy, and then allow suffixation also to be a matter of metonymy, parallel with conversion. Or we deny that conversion is metonymy at all, because derivational affixation is not metonymy and conversion is parallel to derivational affixation.

2. Methodology: data collection

2.1. Using lexicographic resources (instead of corpora)

28When choosing linguistic data as the basis for analysis, researchers have lately been drawn towards broad corpora. There are numerous advantages to using corpora, among others: quantity, or variety. As Desagulier [2017: 7] points out “the goal of corpus-linguistics techniques is to better understand the rules governing a language as a whole, or at least some aspect of that language”. This suggests that a corpus could be an appropriate resource for modelling denominal verbs. Indeed, this approach would certainly result in the extraction of verbs which are not institutionalized in dictionaries (slang words, hapax, regional varieties, etc.).

2.2. Extracting and sampling the data

31Data collection resulted in a set of 5 932 denominal verbs attested from 1800 onwards, among which 165 originate from a PN. In the case of polysemous verbs, only the first attested meaning was selected. In the dataset, what is considered a PN is based on the two dictionaries and includes: names of persons (Stalin) or anthropomorphic animals (Scooby Doo), placenames (Manhattan), proprietary names (Zoom), names of groups or organizations (Taliban), and the title of one movie (Gaslight).

33The methodological choice of selecting a random sample of equivalent size for comparison with PN-verbs warrants further details. Considering the size of VdenomEN (5 932 entries), I carried out analyses on a sample of the data. I used basic R functions to randomize the entries. Then, I began the encoding of the variables while monitoring the structure of the sample with respect to the morphological variable. Table 1 gives a comparison of the sample for two different steps in the encoding, namely after the encoding of the first 100 entries, and after the encoding of the first 500.

Table 1. Sample structure comparison of VdenomEN (sample 100 VS sample 500)

34As Table 1 shows, the overall structure of the data does not massively evolve from sample 100 to sample 500. I proceeded with this test every 100 entries. The structure, even though it becomes more refined the more entries are included, appears to remain stable according to the morphological variable.

35The choice of the variable as well as the choice of the granularity of the variable highly influence the results of this test. For instance, if I consider a finer-grained morphological classification, I obtain as many as 47 different morphological patterns (-ize, -ate, de-+-ize, etc.) instead of the 6 above. As a result, other choices could have been made considering the monitored variable and the granularity of the variable. My choice was driven by access and efficiency: the morphological variable is more easily accessible than semantic variable(s); and taking 6 categories into account instead of 47 makes it easier to draw generalizations.

3. Methodology: data encoding

37The Vdenom project involves the encoding of many variables which details prosody, phonology, morphology, semantics and argument selection of both input and output. The present study only focuses on the role of metonymy in verbal meaning. As a result, this section only deals with two variables: morphological processes and metonymic patterns.

3.1. Encoding morphological processes

39The value of encoding morphological processes is hence mostly descriptive and allows to identify if correlations can be drawn between morphological processes – which are usually mapped with certain meanings – and metonymic patterns, whose aim is to encode the semantic shift from base to verb. Plag [2003: 231] explains that the meanings associated with conversion and suffixation usually overlap: “conversion is the most general case in that the meanings of the derivatives with overt suffixes are a subset of the possible meanings of converted verbs”. However, I previously showed (Héois [2020: 104]) that some meanings tend to attract one type of morphology over the other, so that the morphological and semantic variables may partially be redundant.

40For this set of data, I encoded the morphological processes through two types of granularities: coarse- and fine-grained. Table 2 below shows these two types of encoding of the data.

Table 2. Morphological encoding of denominal verbs

41Most of the processes in Table 2 are defined in Section 1.2.1. The “Complex” category includes cases in which two (or theoretically more) morphological processes co-occur, for instance backformation and suffixation. As a result of this encoding, two morphological variables are coded, containing respectively 6 and 19 modalities.

3.2. Encoding metonymies

42As explained above and as can be inferred from Plag [1999] and Héois [2020], the same formal constraints – phonological, prosodic and morphological – appear to guide denominal verbal derivation, regardless of the nature of the base. As a result, these formal variables are not relevant to account for the difference(s) between PN- and CN-verbs. Hence, the present study explores metonymic patterns as a potential explicative variable. In order to prioritize comparisons with other works on metonymies and word-formation, I borrowed most of the source and target terminology from Janda [2011], who herself based her classification on Peirsman & Geeraerts [2006]. Table 3 partially reproduces Janda [2011: 372] but only retains the terminology which is useful here. The nature of the present data is restricted to animate entities, as a result, the sources and targets involved are less diverse than in Janda [2011].

Table 3. Terminology for sources and targets, based on Janda [2011: 372]

Relating to Actions:	action, state, change state, event, manner
Relation to Participants:	agent, product, patient, theme*, result*
Relating to Entities:	entity, characteristic, leader
Relating to Part for whole:	located, location

45For instance, Janda [2011: 373] briefly defines event as “some result from a verb”. Such a definition directly suggests a deverbal process which would be incompatible with my denominal data. The event class, however, appears useful to the analysis of the source/target of a number of entries in the data, as illustrated in (9):

(9a) Verb – Morganize: “To assassinate or murder (a person) so as to prevent or punish disclosures”. [OED 2022]

(9b) Morphology – Formed by ize-suffixation on the PN-base Morgan.

(9c) Base – William Morgan is believed to “have been secretly assassinated by the Freemasons in 1826” [OED 2022]. The definition of the verb implies two possible motives for this murder (“prevent or punish disclosures”).

(9d) Metonymic patterns – patient for event + event for action.

(10) patient: “Undergoer in an event that experiences a change of state, location or condition, that is causally involved or directly affected by other participants, and exists independently of the event”. [VerbNet]

46The notion of event is less restricted here than in Janda’s definition [2011: 372], as shown in (11):

(11) event: something that happens whether voluntarily or not.

However, considering some of her examples (holiday, festival), her own understanding of event may also be broader. An event is spatial and temporal, as is an action (Peirsman & Geeraerts [2006]), but I suggest that it does not necessarily involve an instigator, as is the case of natural events such as tsunamis, earthquakes or heatwaves for instance.

47The source/target action is also defined relatively to time and space, as are state and change state. Janda [2011: 373] defines them as follows and in opposition to one another:

(12) change state for verbs that describe changes of state (as opposed to static states or actions […]); […] action (for verbal actions not classed as state or change state […])

48These three classes cover verbal derivation and are always placed as final targets of the metonymy accompanying verbal derivation. Their implementation to the data may need some clarification. For instance, in (9), while assassinate is in partial synonymy with Morganize and is usually classified as a change of state (Levin & Rappaport Hovav [2005: 73]), the metonymic pattern paired with verbal derivation does not include change state as a target but action instead. In this analysis, the change of state (from alive to dead) is already included in the first metonymy and is part of the core meaning of the event. As this event acts as the source of the verbal metonymy, its core meaning is retained in the derivation while the metonymy describes the action of carrying out the event.

49The last relevant type of source/target relating to actions is manner. It is illustrated in (13). No definition of the notion is given in Janda [2011] or in Peirsman & Geeraerts [2006] from whom she borrows the category. As it is grouped with actions but does not involve verbal derivation, I define it in opposition to event. Indeed, this notion can be understood as a type of event in which the semantic focus lies on the way the event happens rather than on the event itself.

(13a) Verb – spider: “To catch or entrap after the manner of the spider”. [OED 2019]

(13b) Morphology – Formed by conversion of the CN-base spider.

(13c) Base – spider: “One or other of the arachnids belonging to the insectivorous order Araneidæ, many species of which possess the power of spinning webs in which their prey is caught”. [OED 2022]

(13d) Metonymic patterns – agent for manner + manner for action.

In this example, a specific action carried out by spiders, namely <manner to catch a prey> is selected from the noun in order to serve as the core meaning of the verb. As the spider is an animate entity which instigates the action, it is analyzed as the agent. Similarly to patient, agent is a notion borrowed from argument analysis and can be defined as follows:

• 52 The notions of agency, consciousness, and intentionality are understood broadly. As a result, even (...)

(14) Agent: “Actor in an event who initiates and carries out the event intentionally or consciously52, and who exists independently of the event”. [VerbNet]

50While it could be considered a type of event, the manner class is relevant because it gives information on the core meaning of the verb. As a result, I propose the following definition for the sake of this study:

(15) Manner: the way in which something happens or is done.

51Among the classes relating to participants, I have already given a definition of agent and patient. Product is also a common thematic role as defined in argument structure studies and often defined as a “[r]esult that is a concrete object” [VerbNet]. However, Janda gives a slightly different definition [2011: 373]:

products are created in the context of the metonymy relationship described. Thus Czech sbírka ‘collection’ has a product as target since the collection did not exist prior to the collecting.

(16) Product: result of an action or process whose existence is dependent on these actions or processes.

52Consequently, scientific processes such as the ones invented by Pasteur (pasteurize: “To subject (milk, wine, food, etc.) to pasteurization” [OED 2022]) or McAdam (macadamize: “to make or repair (a road) according to McAdam’s method” [OED 2022]), among others, are coded as product, as well as other cognitive processes related to politics (de-Stalinize: “To counteract the excesses of Stalinism or the influence of Stalin” [OED 2018]), psychology (Pelmanize: “To practise Pelmanism” [OED 2022]) or philosophy (plotinize: “To philosophize in the manner of or according to the principles of Plotinus [OED 2019]).

(17) Pattinsonize: “To remove silver from (lead) by Pattinson’s process” [OED 2022] >> agent for product.

(18) Grimthorpe: “To restore (an ancient building) with lavish expenditure rather than skill and fine taste” [OED 2018] >> agent for manner.

(19) Mithridatize: “To render immune to or tolerant of a poison, esp. by the administration of gradually increasing doses” [OED 2022] >> agent for product.

55In all the examples above, the source category is either agent or patient. This encodes a specific relationship with the target of the metonymy. For instance, the product target needs an agent who created the product as its source. However, the link between source and target can be different, especially when an inherent characteristic of the source is selected. Because a characteristic is considered to be part of the animate entity (either physical or psychological characteristic), the relationship between source and target is a specific case of contiguity, namely part/whole. As a result, the source can also be an entity from which an inherent characteristic is selected. Physical and psychological characteristics are often considered inherent to the entity however, a part of subjectivity is unavoidable to decide whether the semantic element is a characteristic or an event.

(20a) Verb – Poodle: “To clip or shave so as to resemble a poodle”. [OED 2022]

(20b) Morphology – Formed by conversion of the CN poodle.

(20c) Base – Poodle: “A breed of dog, of which there are several varieties, with tightly curling hair, usually black or white and often ornamentally clipped or shaved”. [OED 2022]

(20d) Metonymic patterns – entity for characteristic + characteristic for change state.

56Example (20) illustrates three source/target elements which have not yet been exemplified. First, the final target is change state. In (9), I explained that in Morganize the feature <change of state> is already included in the event. The case is different in (20) as the specific characteristic of the base which is selected can be paraphrased as <ornamentally clipped or shaved> or more simply <prototypical poodle appearance>. This feature does not involve any preliminary process which would result in this appearance, as is expected for inherent characteristics. As a result, when the characteristic is turned into a verb through the second metonymic pattern, the notion of change of state – which is present in the verb definition – needs to appear in the target.

57Neither Janda [2011] nor Peirsman & Geeraerts [2006] define what entity means. From various examples in Janda [2011: 374, 379], such as Kafka, olive tree, steamboat or tough guy, I inferred that a general definition of the term was suitable:

• 56 As pointed out by an anonymous reviewer, the adjective “real” is problematic here as fictitious obj (...)

(21) entity: “concrete. Something that has a real existence56 […] as distinguished from a mere function, attribute, relation, etc.”. [OED 2020]

58The issue is similar for characteristic as the definition is missing. Once again, judging from the examples in Janda [2011: 379-380], such as horned, tough, tasty, I selected the OED definition:

(22) characteristic: “A distinctive mark, trait, or feature that may serve for identification; a distinguishing or essential peculiarity or quality”. [OED 2022]

59Classes (21) and (22) are examples of categories relating to entities. The last of these categories, which is relevant for this study, is leader. Janda [2011: 373] defines it as follows:

(23) leader: “specific type of entity”.

This short definition is made clearer through the example she gives: “Czech hitlerovec, ‘follower of Hitler’ derived from Hitler is an example of leader for entity” (Janda [2011: 373]). This illustrates the usefulness of subcategorizing entity in this case: here, the leader terminology makes it possible to signify the asymmetrical relationship between source and target, while entity1 for entity2 would not be as clear.

60The last category in Table 3, namely classes relating to part for whole, is quite different from the previous classes, as they function as pairs and exist because we often conceptualize entities in their relation to other contiguous elements through a part for whole or whole for part relationship. The located/location pair is a subtype of this part/whole relationship as explained by Peirsman & Geeraerts [2006: 280-281]:

[A] container is conceptualized as a (functional, almost experiential) whole, on the basis of the containment relation with its content, its part. […] The precise relation between part & whole and container & contained seems to be that of a continuum that can be described in terms of ‘‘strength of contact’’. […] If we now allow this ‘‘strength of contact’’ to become a bit looser still, we arrive at a third metonymical pattern from our inventory: location & located. In these metonymies, an entity is referred to by its location or vice versa.

In other words, the located for location pattern is conceptualized as a part for whole pattern because a location can be conceptualized as a container. The advantage of taking these patterns into account is similar to the argumentation for leader above: while both located and location are subtypes of entities, the part/whole relationship they share is lost in the pattern entity1 for entity2. (24) gives a definition of the two classes:

(24) located/location: an entity is conceptualized through its spatial relationship to the place (location) in which it is situated (located).

61In order to better accommodate my data, I chose to add two classes to Janda’s classification: theme and result. Both are types of participants and borrowed from syntactic theory:

(25) theme: “Undergoer that is central to an event or state that does not have control over the way the event occurs, is not structurally changed by the event”. [VerbNet]

Contrary to the patient, the theme is not affected by the action. Example (26) illustrates it:

(26a) Verb – Idiot: “To call (a person) ‘idiot’”. [OED 2019]

(26b) Morphology – Formed by the conversion of CN idiot.

(26c) Base – Idiot: “A person without learning; an ignorant, uneducated person”. [OED 2022]

(26d) Metonymic pattern – theme for action.

62The notion of result is close to product and can be conceptualized as a subtype. Similarly, it is borrowed from syntactic analysis and can be defined as follows:

(27) result: “Goal that comes into existence through the event”. [VerbNet]

The notion of “event” here is understood broadly as example (28) illustrates:

(28a) Verb – Roentgenize: “To reveal as though by subjecting to X-ray analysis”. [OED 2022]

(28b) Morphology – Formed by ize-suffixation on the PN-base Röntgen.

(28c) Base – Wilhelm Conrad Röntgen (1845–1923), German physicist, who discovered X-rays in 1895. [OED 2022]

(28d) Metonymic patterns – agent for product + product for result + result for action.

In (28), the invention (product) is selected from the scientist’s name but, contrary to Pattinsonize in (17), the subsequent action is not only the application of the product (<subject to X-ray>) but more specifically the result of said product (<reveal>).

63As mentioned above, there is always a part of subjectivity in any classification and other categorizations might yield other results. The definitions above make it possible to reproduce the present analysis.

4. Data description and analysis

4.1. Structure of the data

64Out of 194 entries, 100 (51.5%) originate from a PN and 94 (48.5%) originate from a CN. Figure 1 shows that animal referents are few and far between, and almost inexistent for PN-verbs (scoob is the only PN-entry).

65As explained in Section 3, the morphological variable is relevant to describe the data and compare it to other data – namely the broad denominal verb database as introduced in Section 2. Figure 2 presents the structure of the data according to the coarse-grained morphology of the verb. The figure is divided according to the type of the base and the colored columns make it possible to visualize the finer-grained morphology.

68Consequently, I argue that the difference in the morphological structure between PN-verbs and CN-verbs, namely the over-representation of ize-suffixations in the PN data, is not linked to a fundamental difference between PNs and CNs, but to the over-representation of scientific jargon in lexicographic resources. The argument is as follows: person PNs are spatially and temporally situated. As a result, most coinages on this base are hapaxes or occasionalisms and very few become institutionalized. However, verbs built on names of scientists are usually coined within their specialty field and used to denote the action of performing a new process, or using a new invention (pasteurize, for instance). Consequently, their use tend to be institutionalized more easily and they generally become obsolete or evolve only when the process or invention is outdated. So I consider this difference as a methodological limit linked to the choice I made to extract the data from lexicographic resources, and not the symptom of a fundamental difference between PNs and CNs.

4.2. Proper name-to-verb derivation: a two-step cognitive process

70The encoding of metonymic patterns shows three types of verbs. Some verbs only involve a single pattern, some involve two patterns, and a last group involves three subsequent patterns. When the derivation only involves one metonymic pattern, the metonymy directly leads from the base to the verb. In the case of two or more patterns, I hypothesize unknown categories X and Y which can, but need not, be nominal. As an illustration, I reuse example (1) which is reproduced in (29):

(29a) Verb – Bogart: “to force, coerce; to bully, intimidate”. [OED 2021]

(29b) Morphology – Formed by conversion of the PN Bogart.

(29c) Base – Humphrey Bogart (1899–1957), U.S. film actor who sometimes played “tough characters”. [OED 2021]

(29d) Metonymic patterns – agent for event + event for characteristic + characteristic for action.

This example involves a three-step metonymic process in which the intermediary categories – for event and characteristic – are not specified.

71Figure 4 presents a visualization of the data according to the number of metonymic patterns involved in the derivation:

5. Discussion: what verbal derivation brings to the definition of proper names

74The data under scrutiny and its analysis confirm the hypothesis that the nature of the base, whether it is a PN or a CN, determines the number of metonymic processes at work in verbal derivation. The question whether the reason behind this difference lies in the definition of PNs is far from being resolved, however. A closer look at Figure 4 above may be relevant as it shows that a sub-group of converted CN-verbs also undergoes 2 metonymic processes during verbal derivation. This sub-group appears to gather most of CNs which denote animals. A Chi² test and a residual test show that the “animal” variable is significantly linked to the number of metonymic processes involved in derivation, as shown in Figure 6 below.

The difference between animal and person referents for CNs is thus significant in understanding their verbal derivation. Person-CNs undergo a single metonymic process while animal-CNs tend to undergo a double process.

75Considering the data is restricted to very few types of nominals, other subcategories in both CNs and PNs may show one or the other derivational profile and further research is needed. The common denominator between animal-CNs and PNs appears to lie in the underspecification of the referent. In other words, verbal derivation only builds on specificity. As a result, when the core meaning of a base is too broad, specific elements need to be selected to coin the new verb. Indeed, when a coiner considers the nominals spider or Bogart as candidates for verbal derivation, they can choose among a wide variety of elements to build the meaning of the verb. Many of these features could be considered salient: a spider can be noticed for its hunting capacities and techniques, but also for its shape (characteristic), its diet (characteristic), its habitat (location), the way it moves (manner), the feelings it provokes in humans (event), etc. Similarly, Humphrey Bogart can be remembered for many elements of his life such as his looks (characteristic), the way he behaved with other people (characteristic), famous anecdotes from his life (event), etc. Conversely, person-CNs do not show such a wide array of potential features. For instance, a chef is defined by its core function and only subfeatures of this already specific feature can be selected for derivation, the same goes for instance for bigmouth (“to brag (about)” [GDS 2022]), or childmind (“To look after (a child) as a childminder” [OED 2022]).

76As a result, PNs, as well as animal-CNs, involve at least a two-step cognitive process when they are derived as verbs because of their semantic underspecification. Both types of nominals denote complex multifaceted entities. Further research is needed – and planned – to explore the types of metonymies involved in derivation and discuss the notion of saliency which is probably relevant in feature selection.

78However, a closer look at CN-verbs shows that animal-CNs behave more like PNs than person-CNs in verbal derivation as they appear to “need” a two-step cognitive process as well. Indeed, specific features of animal-CNs are necessarily selected before the verbal derivation is possible. This questions the nature of the difference between CNs and PNs in verbal derivation as some CNs appear to behave differently. As a result, while Philippe’s claim may apply here to explain, at least partially, the difference between CNs and PNs in that PNs “are not nouns, [but] only move toward the noun and draw their inspiration from the semantics, morphology and syntax of nouns”, it is doubtful that the same claim would apply to animal-CNs. The type of the base is thus relevant in verbal derivation but the distinction between PNs and CNs is insufficient, and a finer-grained categorization is needed to account for verbal derivation. Beyond nominal definitions, this study suggests that denominal verbal derivation involves semantic constraints: only specific semantic features can serve as a base for derivation.