An Upper Take on Doubler-Uppers

Abstract

Against the background of comparatively insufficient, expressly dedicated studies on double particle verb person nominalisations, this paper offers a qualitative, cognitive-constructionist approach to the properties of doubler-upper nominalisations of particle verbs in English and a reappraisal of some of the available analyses thereof. On the assumption of the validity of the flexicon stance on the organisation of words in the human mind, and on the basis of a preliminary semantic analysis of 300 types of doubler-upper nouns extracted from two corpora and Urban Dictionary, it is claimed that there are no identifiable constraints on the possibility of double-er marking and no particular properties of particle verbs as bases to preclude double -er marking. A hypothesis is formulated that, despite their deviance, doubler-uppers strike the optimal balance between complexity and unity and appear to be the most natural and morphophonologically best-fitting pattern for particle verb -er nominalisation (at least in spoken discourse and the media).

1. Introduction

Doubler-upper nouns (Cappelle’s (2010) label) illustrate two of the “fundamental problems of derivational semantics”, as identified by Kotowski and Plag (2023, p. 2), ensuing from the nature of form-meaning mappings—“polysemy” and “form-meaning mismatches”. As one of the competitors in a quadruple-wide1 competition for the person nominalisation of multiword verbs in English, doubly -er marked de-particle verb2 nouns display polysemy in terms of possible readings, including agent (e.g., asker-outer), stimulus (e.g., calmer downer), entity (e.g., fixer-upper—with a potential patient reading as well, besides the agentive one), and event (e.g., cutter-offer, backer-outer), while the seemingly semantically empty doubling of the affix shows “form-meaning mismatches”.

Doubler-upper nouns raise uneasy questions with their mere identification—are they affixed phrasal compounds, reduplicative compounds, or affixed complex lexemes? They cannot be easily classified as phrasal compounds, such as do-it-yourselfer, since, according to the discussion in Bauer et al. (2013, pp. 509–17), suffixation on compound and phrasal bases occurs at the outer rightmost boundary (irrespective of whether precise differentiation can be established between a compound and a phrase), while, as the label iconically suggests, they display two instances of -er suffixation, one of which disrupts internal (phrasal) unity. Their affixed wordhood is not immediately or easily evident, even when Haspelmath’s (2023) newest definitions of word (as a text element and as a lexeme) are adopted. Multi-word expressions (MWEs) encompass compounds, complex verbs, and idioms (Schulte im Walde and Smolka 2020, p. iii) or “linguistic objects formed by two or more words that behave like a ‘unit’ by displaying formal and/or functional idiosyncratic properties with respect to free word combinations”, comprising “an extremely varied set of items (from idioms to collocations, from formulae to sayings)” (Masini 2019, p. 1). Subsuming doubler-uppers under MWEs resolves the issue of their participation as derivational bases in word formation processes, but leaves the problem with their double suffixal marking (where one of the instances of the suffix seems vacuous) open.

Atheoretically, we can use at least the following labels, each of which emphasises different properties of such nouns: reduplicative compound, agentively affixed multi-word verb (MWV), doubly affixed de-particle verb nouns, (constructional) template, and the product of non-semantic affix doubling, etc. The label ‘noun’ presupposes the following: it is a word with an identifiable part of speech membership, it allows pluralisation, and it can potentially perform the syntactic functions of nouns such as boy or stone. As nouns, doubler-uppers do not easily fit into the derivational ecosystem of English, as they are non-typical nominalisations (Lieber 2016), since coindexation with two separate bases of the same affix (if it is affixal reduplication) is hard to model. The multi-word verb itself has already been coindexically integrated into a single meaningful unit combining a verb and a particle, which can further be suffixally augmented, but typically with a single suffix.

The analysis offered below takes stock of these issues in relation to doubler-upper nouns and puts forward a cognitive-constructionist approach, which provides a few specific answers to some of the questions hovering around these nominalisations. The paper is structured as follows: Section 2 revises the nature of doubler-upper nouns, arguing against a reduplicative interpretation; in Section 3, a cognitivist-constructionist alternative is offered, and Section 4 concludes.

2. The Nature of Doubler-Uppers

2.1. A Preliminary Constructional View of Doubler-Uppers

Within constructionist analyses (e.g., Audring 2019; Booij 2016, 2017, 2019; Booij and Masini 2015; Goldberg 2006, 2019; Hilpert 2015, 2019a, 2019b; Hoffmann 2022; Mansfield 2021), the construction type is considered to be significant, and words, but not morphemes, are considered to be the smallest constructions in terms of complexity in construction morphology (Booij 2013, p. 256), while morphemes are recognised as constructions in construction grammar (Goldberg 2006, p. 5). At least the following construction types of a word in terms of complexity with distinct properties have been recognized, “where ‘M’ is a morpheme, “[_]w” is a constituent that meets wordhood criteria, and ‘+’ indicates one or more iterations of an element type.

[M⁺]w = (Minimal) Word

[[M⁺]w -M⁺]w = (Recursive) Word

[[M⁺]w [M⁺]w⁺]p = Phrase”

(Mansfield 2021, p. 1431)

In view of these considerations, the bases of double-uppers are either recursive words or phrases further participating as recursive words in derivational processes by combining with morphemes. Irrespective of their recognition as words or multi-word semantically unified expressions, they participate in at least one morphological schema in the sense of Booij (2013), which is associated with the semantics of the ‘agent of the activity named by the unified concept of the base’.

Another set of properties of construction types captures the nature of the internal constituents in terms of fixedness and substantive specificity or variability, where a constructional idiom or an abstract schema is recognised as one in which at least one constituent is phonetically specified, e.g., “<[x]_Vi er]_Nj [Agent of SEMi]j>” (Booij 2016, p. 425). Even if we recognise the preservation of the mapping with the semantic part, for doubler-uppers, we need to adequately model the formal part that will encode the agent (or other extended) reading(s), either despite the double appearance of the suffix or supported by it.

In other words, doubler-uppers are not typical compounds for at least the following two considerations: (i) they are considered to arise either through reduplication or affixation (by no means coterminous or easily reconcilable processes in the derivational ecosystem of English); and (ii) their bases are not typical compounds (variability in ordering, which leads to internal interruptability, which individually and conjoinedly undermine wordhood). They cannot be considered as affixed phrasal compounds either, since, in phrasal compounding, any affixation would be at the outer rightmost border. The doubling of the affix seems not to be semantically motivated, i.e., they are not typical affixal products. Reduplication in English seems to be naturally associated with compounding (Benczes 2012), not with affixation, and is considered “not a major means of creating lexemes in English, but […] perhaps the most unusual one” (Crystal 2003, p. 130).

From a construction-analytical perspective, issues are raised relating to their identification as alternating schemas inherited from a single mother schema, as constructional idioms (with double -er substantively specified) or constructional templates.3 They could be identified as constructional templates in Hoffmann’s sense (2022, p. 14), but only if they are assumed to be recursively derived, and not if they are marked simultaneously (see the interpretation of great-great-grandfather in Hoffmann (2022, p. 14)), so they do not fit the criteria for a constructional template either. The term template in relation to affixal elements also raises other associations—a pattern of ordering without a hierarchical structure (Mithun 2016, pp. 149–50) and fixed slots for particular meanings. Any which way an analyst looks doubler-uppers seems to be the odd man out.

2.2. What Has Been Established about Doubler-Uppers

Both multi-word verbs and -er deverbal nominalisations in English have been extensively studied from various perspectives (e.g., Baeskow 2015; Bolinger 1971; Heyvaert 2010; Lieber 2004, 2016; Lieber and Andreou 2018; Ryder 1999, 2000). Their mutual products have not been commensurately studied, with a few full-length studies exclusively dedicated to the problems these pose for numerous widely shared assumptions (see Cappelle 2010; Chapman 2008; Denison 2008; Lensch 2018, 2022; McIntyre 2004; Walker 2009) about word formation.

The synchrony and diachrony of doubler-upper nouns reported in the literature can be generalised into the following:

• General: They are still considered to be marked—either colloquial (Blevins 2006), humorous (Bauer et al. 2013), jokingly nonstandard (Quirk et al. 1985, p. 1539), extravagant (Lensch 2022), “popular, grotesque“ (Wentworth 1936), clumsy (Bauer 1983), intentional errors (Cappelle 2010), or “fairly natural kind of ‘error’” (McIntyre 2004). Supposedly, they are more widely spread in American English (including the written medium, see Lensch 2022), but certain types (roller-upper, washer-upper, roper-inner, and stopper-inner) are attested only in BNC, not in COCA (Bauer et al. 2013, p. 218).
• Diachronic perspective

  (i)

  Chapman (2008) speculates on their historical development, suggesting that the picker-upper pattern may be a transitional stage on the way from picker-up to pick-upper.

  (ii)

  They are a borderline phenomenon, lying “on the chronological or conceptual borders of lexicalization or grammaticalization” (Ryder 2000, p. 292). The suffix -er itself is on the way to full grammaticalization, acquiring a clitic-like character in its ability to attach to “non-lexicalized phrases”4 (Ryder 2000, p. 292). Multiword verbs do not belong to the latter group as they are fully lexicalised, but they seem to occupy a middle ground in the path to grammaticalization between the prototypical deverbal -er nominalising suffix and the clitic-like, anaphoric-function-performing suffix -er.

  (iii)

  They are a transient fashion (Mencken 1945).

• Synchronic perspective

  (i)

  They are a morphological phenomenon (which has to be recognised as a deviant one in the sense of Hathout and Namer (2014)). According to Lensch (2022, p. 76), they violate at least four principles of English morphology: the Righthand Head Rule, the monosuffix constraint, Aronoff’s (1976) Unitary Base Hypothesis, and the avoidance of formally (near-) identical and (near-) adjacent (non-coordinate) grammatical elements or structures. In her view, they are realised by affix reduplication, which is not characteristic of the ecosystem of English word formation (unlike the recognised role of reduplication in rhyming compounds of all types (Benczes 2012; Rastall 2004));

  (ii)

  They are a phonological phenomenon, or at least a phonologically motivated phenomenon—according to Walker (2009), they arise for euphonic causes; in the view of Denison (2008), they are rhythmic phenomena, associated with the specific rhythmic pattern of phrasal verbs. Doubler-uppers, in his view, are “a new rhythmic template, one in which the relative weights of verb stem and particle are respected as well as the need to have -er as final element of an agent noun (which is both a rhythmic and a formal matter)” (Denison 2008, pp. 212–13).

  (iii)

  They participate in a tripartite competition among patterns: V + P-er, V-er + P, and V-er + P-er, and the criteria for discriminating which one is chosen for a particular nominalisation remain obscure (Chapman 2008).

  (iv)

  They violate two principles or strong tendencies in English morphology: (1) the -er suffix has a tendency to attach to verbal bases, and (2) English derivational morphology is normally placed at the right-hand margin of the base (Chapman 2008, p. 279; Cappelle 2010, p. 360). We can add to this the general principle of the right-headedness of compounds in English.

  (v)

  They allow multiple -er attachments at the right outside boundary, e.g., washer-upperer, thus violating Bauer’s contention that “in English no suffix can be added to a base that already ends in the same suffix” (Bauer 1983, p. 92).

To these points, it might be relevant to add that doubler-uppers do not easily or conveniently fit into any of the four types of multiple exponence (ME) identified by Harris (2017). They look like Type 2, the alternating one “typically involving featurally identical non-adjacent affixes”, which is optional “or it may alternate with a simpler form, or it may be inconsistent within a paradigm” (Harris 2017, p. 54). Even though doubler-uppers conform to the formal side of Type 2 and are also lexically governed, they contradict the meaning side of Type 2 ME, since it “is probably always inflectional, including marking of number, case, possessives in the noun, and agreement markers in the verb” (ibid.), while -er doubling on multi-word verbs in English, at least for the time being, predominantly serves naming functions of agent, instrument, stimulus, entity, and event readings (though not place or inhabitant readings, otherwise characteristic of the suffix -er).

Even the recognition of the ubiquity of -er suffixation on all kinds of linear elements, including non-lexicalised phrases, as in in-the-parker, up-and-comer, and blue-chipper (Ryder 1999, p. 292) (where its function is interpreted as an anaphoric marker with generalised substantivizing meaning akin to one), does not solve or directly relate to the doubling issue. On the contrary, it makes it even more special in view of the fact that -er at the rightmost border progressively widens the scope of bases that easily tolerate it. Besides not solving the doubling issue, this invites the recognition of an internal border within an MWE, which, for certain purposes, is a unified linearisation (onomasiologically speaking, MWVs constitute a single naming unit), but for other purposes, displays an internal boundary, e.g., pronominal direct object movement with phrasal verbs (e.g., fix the house up vs. fix up the house and fixer upper with at least two possible readings—agent and entity).

2.3. Why Doubler-Uppers Are Not the Product of Reduplication

As early as the 1930s (Wentworth 1936), doubler-uppers were recognised as affixed compound nouns, which does not run contrary to the principles of English word formation, as many compounds in the language are of the complex type (see e.g., Bauer 2017; Hilpert 2015; Dirven and Verspoor 2004, etc.), i.e., both the processes of compounding and affixation are involved in their constitution—e.g., good-looking, blue-eyed, and hand-carved, etc. From this perspective, doubler-uppers deviate in terms of two-fold marking with the same affix. For Lensch (2018, p. 163), the explanation lies in the fact that “doubler-upper nouns constitute an additional systematic pattern to the classes of reduplicatives in English”. However, doubler-uppers cannot straightforwardly be explained as products of reduplication. They are far removed from both canonical and prototypical reduplication within “the unclear phenomenology of reduplication” (Stolz 2018, p. 201) cross-linguistically. Haugen (2015), in acknowledging the legitimacy of affixes as reduplicants under the Morphological Doubling Theory of reduplication (Inkelas and Zoll 2005), emphasises that “the question of whether or not affixes can be legitimate targets for reduplication” (Haugen 2015, p. 12) remains open, as well as the issues of which affixes can be reduplicants and why these specific affixes.

A reduplicative interpretation of doubler-uppers violates Hurch’s definition of reduplication as a morphological procedure “by which the inflectional and/or derivational formatives used to signal a specific category are directly derivable from the phonological/prosodic structure of the uninflected or underived simplex form” (Hurch 2005, p. 1). Furthermore, Lensch (2018, p. 160) claims that repetition “triggers a change in expressive or interpersonal meaning”, while, as a morphological procedure, reduplication “is accompanied by changes in descriptive meaning”. It also creates bracketing problems and presupposes an intermediary stage in the derivational process. If we are to assume that doubler-uppers result from the morphological procedure of reduplication, we need to assume one of the single-affixed alternatives as the base for reduplication and, thus, acknowledge at least two stages involved in the derivation of these complex nominals—affixation (for the single affix marking), followed by reduplication, e.g., ask out > asker out > asker outer. Again, the doubling remains an unjustified addition, since the target semantics is already present in the single-affixed form, while Booij (2010, p. 39) contends that “the doubled structure conveys a meaning that is not reducible to the meaning of its constituents” (Booij 2010, p. 39). The fact that doubler-uppers are competitors in a tripartite competition suggests that their meaning is reducible to that of any of the intermediary bases (see asker out above).

Lensch (2022, p. 96) explains this away as extravagance and creativity on the part of speakers and emphasises that the double marking has exclusively expressive functions and does not affect the descriptive meaning of the derived noun. In view of the principle of economy and the type frequency of doubler-uppers or the fact that “attestations of this template have become too numerous” (Lensch 2022, p. 90) and their passing “the threshold […] to the written medium” (Lensch 2022, p. 90), such an explanation falls thin. Furthermore, according to Denison (2008, p. 215) “[d]ouble -er does not seem explicable as emphatic reinforcement, as some kinds of linguistic doubling certainly are”. Even if we adopt the definition of discontinuous reduplication, “where other morphological material may appear between the reduplicant and the base” (Velupillai 2012, p. 101), which, according to Mattiola and Masini (2022, p. 274), is both non-canonical and non-prototypical, we still have not solved the issue with the functional justification and meaning-encoding problem of double -er marking. Adopting Good’s (2016) definition of template, Lensch (2022, p. 76) avoids ‘the reduplication as a morphological procedure’ issue by identifying doubler-uppers as templates, since “templates are more flexible than schemas in the Construction Grammar sense”. A reduplicative interpretation dissociates the three alternative nominalising patterns of particle verbs and precludes the possibility of a unified analysis capturing the properties of all competing alternatives and showcasing the common core. It runs counter to the ideas of constructionism and violates both inheritance relations (from a common, more abstract schema to more substantively specified variants) and sister relations (the complex relations between equally substantively specified schemas at the same level of abstraction). It leads to the conclusion that the single marking of the -er affix (no matter where exactly it is positioned—internally on the verb, or externally on the particle) results from a different process, i.e., affixation, while double marking results from reduplication. Reduplication, as a morphological procedure, requires the presence of the reduplicant in the base, which, as explained above for doubler-uppers, means that a vacuous element is added that is an exact phonological copy of an element from the source, but without any morpho-semantic contribution, which is questionable in view of at least two principles—the economy principle and the cognitivist principle that “all structures in language, ranging from morphemes, to words to syntactic patterns, are considered as inherently meaningful” (Lemmens 2015, p. 90).

3. Doubler-Uppers as Sister Allostructions

In other words, doubler-uppers remain a unique type of complex -er nominal, defined by Ryder (1999, p. 292) as “a noun with the nominalizing -er suffix whose base consists of more than one free morpheme”. They exemplify ““deviant word formation”, that is Word Formation (WF) departing from the canonical one-to-one correspondence between form and meaning, and thus moving away from morpho-semantic transparency” (Hathout and Namer 2014, p. 177). This deviance runs contrary to their optimality, paradoxically also defined in terms of the best morphophonological fit.

In Hathout and Namer’s classification of deviant morphology, doubler-uppers are simultaneously an instance of “derivational over-marking” and of “lexical over-marking”. From the point of view of derivational canonicity, the double (or in some cases multiple) marking of -er displays a mismatch between meaning and form correspondence in reference to the local base-derivative relations, while the fact that they participate in a tripartite competition between patterns with the same semantics makes them lexically deviant.

A fitting model to analyse the phenomenon is to conceive a family of constructions or sister-schemas that share a common core but display individual specificities. This may avoid the issue with the exact identification of the morphological process or procedure (reduplication, affixation, and compounding, etc.) of their constitution on the one hand, and on the other, emphasise the affinity among the three alternative patterns for nominalising particle verbs in English. To avoid looking at those as epiphenomenal patterns arising from different processes or involving individual schematic representations, an allostructional analysis is offered, with a focus on the complementary analytical utility of recognising both a shared mother schema and a series of sister schemas, which better illustrates the nature of the constructicon as a highly dynamic and exclusively relational network, matched by the dynamics of the flexicon in the mind.

3.1. Data

For both rhetorical and analytical purposes, this section starts with a long quotation from the crowd-sourced online Urban Dictionary (1999–2024). This may raise a few brows, but “UD captures what most traditional English dictionaries fall short of: recording ephemeral quotidian spoken language and representing popular views of meaning” (Cotter and Damaso 2007, p. i). In addition, Chapman (2008) expressed a desideratum for looking into corpora with a greater representation of colloquialisms for establishing a firmer grasp on doubler-uppers. The elaborate doubler-upper example in (1) demonstrates a complex naming unit with an embedded particle verb nominalisation:

(1)

Story interruptor one-upper adder-toer (UD) (2010)—21 positive votes—1 negative vote

A person that interrupts you mid-story. But then not only interrupts your story, they feel the need to one-up your story with an even better story of their own. And to top it off, they not only interrupt you in the first place, one-up you with their story, they then feel the need to add to your original story they interrupted.

I was in class the other day talking about how my dad went to Haiti to help the earthquake victims, and before I could even finish Frank interrupted me and said his dad not only flew to Haiti to help the victims, but that he also donated a million dollars to the relief act. He went on to say that his dad told him my dad was down there but wasn’t actually doing anything to help. That fucking Frank is such a story interruptor one-upper adder-toer!

Example (1) is an excellent exemplar to showcase that there is no identifiable linguistic factor that can determine the choice of agentive particle verb nominalisation in English (except genre specifications and degree of formality). It illustrates the convictions of cognitive and functional linguists “that virtually everything in language is motivated […] even if very little is strictly predictable” (Langacker 2008, p. 14; emphasis in the original). Both the verb and the preposition are affixed in adder-toer, contra Denison’s (2008, p. 224) claim “for a strong tendency of those particles/prepositions that occur more typically in phrasal verbs to be more typical of the picker-upper pattern and those that are more typical of prepositional verbs to be typical of the picker-up pattern.” It appears in the adjacency of another nominalisation of a (converted) particle verb, in which only the particle is affixed. The example is provided by a single speaker, which may be taken to suggest that entrenchment (for an elaboration of the role and nature of entrenchment in the dynamic language model adopted here, see Schmid (2020)) is item-specific or that disparate factors influence the positioning and number of affixes in MWVs’ (agentive) nominalisations. Both nominalisations have strictly agentive readings in the immediate context and participate in a single naming unit; neither denotes a purposeful or rather a habitual doer of an activity, which runs counter to Lensch’s (2018, p. 178) claim of “a tendency for doubler-upper nouns to denote volitional agents or instruments that are used purposefully and not by accident or chance” and have divergent marking, with one of the constituents with a single (particle-attached) -er affix and the other with double affixation.

This is not a typical example and is not used here to argue for the extravagant nature of doubler-uppers, just to alert the reader to the impossibility of predicting which of the three -er marking alternatives will be used for a particular particle verb (agentive) nominalisation. Each of them forms a different pocket of productivity in the local network of -er nominalisations, with the understanding that X-er Y-er is the most colloquial, even though “it just feels right” (Whitman 2010).

The data presented in

Table 1. Types of doubler-uppers with up, down, in, out, off, and away found in NOW, GloWbE, and UD.

	NOW	GloWbE	UD	Types
-upper	115	36	37	162
-downer	13	3	6	20
-innner	8	9	8	21
-outer	50	21	17	76
-awayer	3	1	5	9
-offer	9	2	3	12

below were extracted from the NOW (2023) and GloWbE (2023) corpora and UD, with a view of their potential for conducting, in the future, an analysis across varieties, registers, and genres. The data sources were chosen based on the possibility for more colloquial data to be collected across genres and varieties (although, at this stage, the differences among varieties and genres have not been analysed). GloWbE encompasses different varieties of English and amounts to 1.9 billion words. The NOW corpus, the largest corpus at 18.6 billion words, was chosen for its contemporaneity and the data it represents—web-based newspapers and magazines—while UD was chosen because it is crowdsourced and corresponds to Chapman’s (2008) advice for more colloquial data. UD was manually searched for each of the target forms. The corpora were searched with the string *-*Y-er, where Y is one of the particles for the respective particle verbs, following the approach employed by Hilpert (2015), and then manually cleaned. The hits option was set to 500 hits for each of the searched strings. Only type counts are presented and they are not capitalised on further in the analysis, neither in terms of frequency, nor in terms of genre and register specification, due to the objective and scope of the current research. The extracted types have an illustrative purpose here, with their in-depth analysis a natural next step. All the extracted types are provided in Appendix A, where the six tables (

Table A1. -upper.

NOW	GloWbE	UD
162 unique types out of 188 overall count
115 types	36 types	37 types
8 types coincide across the three data sources; 11 types coincide between NOW and UD; 7 types coincide between NOW and GloWbE
backer-upper	bar-propper-upper	adder-upper
ball-picker-upper	boomer-upper	backseat washer-upper
beater-upper	breaker-upper	beater-upper
blower-upper	builder-upper	blower upper
breaker upper	cheerer-upper	border-upper
brusher-upper	cleaner-upper	checker-upper-onner
builder-upper	doer-upper	cheerer-upper
butter-upper	dryer-upper	cleaner-upper
caller-upper	filler-upper	cracker-upper
catcher-upper	fixer-upper	daisy pusher-upper
chatter-upper	idea-thinker-upper	doer-upper
cheerer-upper	jumper-upper	fixer-upper
chief-washer-upper	knocker-upper	flipper-upper
cleaner-upper	letter-maker-upper	fucker-upper
cocker-upper	liner-upper	hanger-upper
cover-upper	maker-upper	holder-upper
cracker-upper	masher-upper	hooker-upper
curler-upper	mind-maker-upper	kicker-upper
disher-upper	mixer-upper	knocker-upper
doer-upper	mopper-upper	maker-upper
dog-stink-cleaner-upper	number-maker-upper	messer-upper
doggy-poo-picker-upper	picker-upper	mixer upper
drawer-upper	puke-mopper-upper	necker-upper
dreamer-upper	puker-picker-upper	picker-upper
dresser-upper	putter-upper	ringer-upper
feeler-upper	quicker-picker-upper	runner-upper
filler-upper	ramper-upper	screwer-upper
fixer-upper	rounder-upper	shutter-upper
follier-upper	singer-upper	sneaker-upper
follower-upper	snapper-upper	taper-upper
fresher-upper	stander-upper	thicker-picker-upper
fronter-upper	starter-upper	trader-upper
fucker-upper	stitcher-upper	trash-picker-upper
gammon-winder-upper	waker-upper	trumpster pucker-upper
giver-upper	warmer-upper	upper topper flopper stopper
hanger-upper	washer-upper	waker-upper
hangover-fixer-upper		word maker-upper
holder-upper
holderer-upper
kid-picker-upper
knocker-upper
lifter-upper
litter-picker-upper
looker-upper
loosener-upper
maker-upper
masher-upper
mega-fixer-upper
messer-upper
mixer-upper
non-cleaner-upper
non-fixer-upper
non-messer-upper
non-stuff-maker-upper
old-house-fixer-upper
opener-upper
owner-upper
padfixer-upper
pants-puller-upper
paster-upper
pepper-upper
perker-upper
picker-upper
pooper-upper
price-pusher-upper
pumper-upper
puncher-upper
pusher-upper
quicker-picker-upper
ripper-upper
roller-upper
saliva-sucker-upper
scooper-upper
scraper-upper
sender-upper
setter-upper
shaker-upper
shower-upper
shutter-upper
smasher-upper
snapper-upper
soaker-upper
sock-picker-upper
sopper-upper
splitter-upper
starter-upper
sticker-upper
stirrer-upper
street-cleaner-upper
sucker-upper
sweeper-upper
switcher-upper
table-cleaner-upper
teamer-upper
tearer-upper
tidier-upper
time-taker-upper
topper-upper
toucher-upper
trader-upper
trash-picker-upper
tree-cleaner-upper
tree-putter-upper
tripper-upper
trouser-holder-upper
urban-trader-upper
user-upper
waker-upper
warmer-upper
washer-upper
weigher-upper
whipper-upper
whooper-upper
winder-upper
word-maker-upper

,

Table A2. -downer.

NOW	GloWbE	UD
20 unique types out of 22 overall count
13 types	3 types	6 types
2 types coincide in NOW and GloWbE
calmer-downer	calmer-downer	fixer-downer
closer-downer	faller-downer	hanger-downers
cutter-downer	tearer-downer	one-upper downer
gawker-shutter-downer		oner-downer
knocker-downer		poster taker downer
pusher-downer		shooter downer
putter-downer
sitter-downer
slower-downer
taker-downer
tearer-downer
wolfer-downer
writer-downer

,

Table A3. -inner.

NOW	GloWeb	UD
21 unique types out of 25 overall count
8 types	9 types	8 types
2 types coincide across the three data sources; 2 types coincide between NOW and GloWbE
breaker-inner	cutter-inner	backer-inner
cutter-inner	filler-inner	butter-inner
daily checker-inner	fitter-inner	caller-inner
filler-inner	header-inner	caver-inner
keeper-inner	locker-inner	filler-inner
putter-inner	putter-inner	rubber-inner
sleeper-inner	shover-inner	sleeper-inner
tucker-inner	sleeper-inner	spinner-inner
	sucker-inner

,

Table A4. -outer.

NOW	GloWbE	UD
76 unique types out of 88 overall count
50 types	21 types	17 types
2 types coincide across the three data sources; 2 types coincide between GloWbE and UD; 2 types coincide between NOW and UD; and 6 types coincide between NOW and GloWbE
acter-outer	banger-outer	asker outer
bin-putter-outer	caller-outer	backer-outer
blotter-outer	churner-outer	banger-outer
blower-outer	diner-outer	blacker-outer
blurter-outer	freaker-outer	eater outerer
booger-pointer-outer	hair-puller-outer	eXer-outer
bummer-outer	helper-outer	fader-outer
caller-outer	hole-pointer-outer	freaker outer
candle-putter-outer	kicker outer	happiness sucker-outer
churner-outer	neck-sticker-outer	leaver-outer
closer-outer	pointer-outer	maker-outer
disher-outer	problem-hasher-outer	plane-checker outer
eater-outer	puller-outer	puller-outer
filler-outer	putter-outer	root beer giver-outer
finder-outer	sacker-outer	trash taker-outer
flamer-outer	seller-outer	worker-outer
getter-outer	sitter-outer	zoomer-outer
giver-outer	stopper-outer
goer-outer	taker-outer
hander-outer	whipper outer
hanger-outer	worker-outer
helper-outer
lasher-outer
opter-outer
pointer-outer
pourer-outer
puller-outer
putter-outer
puzzler-outer
reader-outer
ripper-outer
roller-outer
ruler-outer
runner-outer
seeker-outer
sitter-outer
sniffer-outer
snuffer-outer
sorter-outer
spoon-putter-outer
stomper-outer
striker-outer
taker-outer
thinker-outer
trash-taker-outer
trier-outer
walker-outer
wiper-outer
worker-outer
zoner-outer

,

Table A5. –awayer.

NOW	GloWbE	UD
9 unique types out of 9 overall
3 types	1 type	5 types
No coincidence of types across the data sources
football-snatcher-awayer	padder awayer	runner-awayer
giver-awayer		hair-goer-awayer
thrower-awayer		keeper-awayer
		cab runner awayer
		pusher-awayer

and

Table A6. –offer.

NOW	GloWbE	UD
12 unique types out of 14 overall
9 types	2 types	3 types
2 types coincide between GloWbE and UD
blower-offer	pisser-offer	pisser-offerer
payer-offer	setter offer	shower-offer
pusher-offer		cutter-offer
ripper-offer
suit taker-offer
sicker-peeler-offer
switcher-offer
taker-offer
ticker-offer

) contain the data from the three sources, arranged per affixed particle: -upper (Table A1), -downer (Table A2), -inner (Table A3), -outer (Table A4), -awayer (Table A5), and -offerer (Table A6). They were qualitatively, not quantitatively, explored in search of any conspicuous factor that may suggest any constraints for double -er marking on particle verbs. (For quantitative data on doubler-uppers, see Chapman (2008) and Lensch (2022), even though two considerations should be taken into account: the data in Chapman are outdated at present and those in Lensch (2022) are genre-specific.) The factors considered are: interference from homonymous full words (offer, upper, downer, outer, and inner vs. *awayer), on the hypothesis formulated by Denison that avoidance of homonymous comparative adjectival forms will facilitate doubling; support for symmetrical coercion via antonymy (up vs. down; in vs. out), on the hypothesis that it is likely that the antonymic association might strengthen the production of symmetrical nominalisations on antonymous particles; and figurativity of meaning, adverb particle vs. preposition as constituents of MWVs, and transitivity of the particle verb base.

Assuming the web is a corpus (Gatto 2014; Kilgarriff and Grefenstette 2003), we can use certain examples found by specific searches as evidence against the euphony-rhythmic explanation or motivation for the marking of the -er in particle verbs, since, as attested below, it can be attached to any element with unconstrained flexibility and can be further repeated multiple times, which undermines the rhythmic argument, e.g., as in the examples (2–6) below:

(2)

Maker-goer-awayer (https://literalminded.wordpress.com/2010/03/02/picker-uppers-and-putter-upper-withers/ accessed on 14 November 2023),

(3)

Putter-upper-wither (https://literalminded.wordpress.com/2010/03/02/picker-uppers-and-putter-upper-withers/ accessed on 14 November 2023),

(4)

Light-turner-offer-onerer (https://morph.surrey.ac.uk/index.php/author/tim/ accessed on 13 November 2023),

(5)

Headache-bringer-oner(er) (https://morph.surrey.ac.uk/index.php/author/tim/ accessed on 14 November 2023)

(6)

Singing comer-outer (https://www.queerty.com/andrew-cristi-singing-comer-outer-20100408 accessed on 16 November 2023).

3.2. Analysis

3.2.1. The Analytical Model

Considering that constructionism resolves many significant points of difference between MWEs and word formationally created complex lexical items, modelling multiple -er affixations on MWEs from a cognitive-constructionist perspective seems like an Occam’s razor approach. Agentive -er marking on particle verbs occupies a special place in the constructicon of English, and its flexibility ensues from the mere nature of constructions, understood as “emergent clusters of lossy memory traces that are aligned within our high- (hyper!) dimensional conceptual space on the basis of shared form, function, and contextual dimensions” (Goldberg 2019, p. 7).

The constructionist strand employed combines findings, methodological decisions, and presentation formats from construction grammar—Goldberg (2019), Hoffmann (2022)—and construction morphology—Audring (2019) and Booij (2016, 2019). The cognitivist stance embraces “the methodology of analytic thought, which includes the systematic manipulation of ideas, abstraction, comparison, and reasoning, and which is itself introspective in character” (Talmy 2007, p. xi). Support for the latter comes also from psychodynamic interpretations of the mental lexicon within psycholinguistic approaches to complex words. Libben (2021, p. 1) argues strongly against the view of the mental lexicon as “a static repository of representations with fixed structural properties”, ensuing from the generally accepted, underlying metaphor that the mental lexicon is a dictionary. The scholar offers an alternative hypothesis, construing the mental lexicon as a flexicon, a dynamic and highly interconnected system of actions. Within the flexicon, each word is a set of activities that we perform. These are organised via the principles of morphological transcendence and lexical superstates. A lexical superstate lacks a fixed structure and is a hub “of alternative morphological structures […] as potential realization” (Libben 2021, p. 1). It is not an unfounded conjecture to claim that the flexicon is the psycholinguistic (or mental) counterpart of the dynamic, output-based system of the constructicon. Analytically, the flexicon is the bridge between cognitive and constructionist views on complex words and can be used to convincingly model the dynamics of alternative -er affixation patterns on MWVs.

3.2.2. Speakers’ Self-Reflections

One methodological technique in cognitive analysis is the assessment of the “accessibility” of language phenomena to the second level of consciousness (Talmy 2018, p. 219) of speakers. In this respect, the metacognitive salience of doubler-uppers is high, as can be concluded from various people attending to it offline in Talmy’s (2018, p. 222) sense, e.g., Whitman (2010), Zimmer (2015), Zwicky (2013), and Feist (2019). Spontaneous linguistic reflections on salient language phenomena tap into an individual’s language consciousness. The concurring consensus among several individuals in journal columns and blogs is that, doubler-uppers, though unevenly considered to be acceptable (especially in formal written genres), just sound right.

One issue discussed in popular texts on doubler-uppers is worth commenting on—the effect of double agentive affix marking on the attachment of inflectional markers such as the plural -s. Feist (2019) comments on the positioning of inflectional material on nominalisations of phrasal verbs and recognises two possibilities: internal inflection on the verbal element (e.g., passers-by or the washers-up) or external inflection (e.g., washer-ups), with the second sounding awkward and rarely being used. But if the agentive marker appears on both constituents, then there seems to be no choice, and plurality tends to be marked externally, which suggests that doubly marked nominalisation is perceived as a non-divisible single unit. This only comes to strengthen Cappelle’s (2010, p. 335) verdict that “doubling the -er may be the most felicitous option when deriving nouns from phrasal verbs”.

3.2.3. An Allostructional Family View

As argued for in Section 2.3, doubler-uppers are not extravagant reduplications, rather, they are optimal affixal variants in a competition between a single verb-based attached affix, single particle-attached affix, and double affixation on both the verb base and particle within the MWV. The answer to the question of why double marking seems to be the optimal variant, in my opinion, is because it strikes the perfect balance between unity and complexity and provides the most semantically fitting solution to expressly mark a fully unified and substantivized form. The variant of single outer affixation reduces the complexity of the base and results in clashes with homophonous single lexical items like inner, outer, downer, and upper, which supports the idea that homonymy with independent words (be them comparative adjectival forms or nouns) functions as a bootstrapping mechanism for affix doubling. Single verbal part affixation disrupts the conceptual unity of the whole, although it satisfies the headedness condition as the locus for various morphological processes. The double -er marking iconically emphasises both the complexity and the unity of the base: the sameness of the affix mirrors the unity, while the doubling isomorphically via diagrammatic iconicity highlights the complexity. Recognising doubler-uppers as the result of affixation with affix doubling, not of reduplication, more naturally accommodates the polysemy that is starting to arise with some of these nominalisations, a phenomenon characteristic of the suffix -er in the ecosystem of English nominalisations, as expounded by Lieber (2016), and points to the unity in variable -er suffixation on MWVs.

Doubler-uppers form a local family of allostructions, together with the positionally differentiated single marking within the extended family of -er nominalisations and the wider network of nominalisations in English in general. Figure 1 presents the allostructional sister relations and the secondary mother schema with which they are related and which serves as the hub that establishes network relations with other kinds of relatives within the extended -er family.

The allostructional family is a local “hierarchy of actions” in Libben’s (2021, p. 2) sense and may be conceived of as a volatile superstate of -er nominalisations of any particle verb. When committed to the nominalisation of an MWV, a speaker will most probably choose, on different occasions, one or another of the available options. The mother schema (on top in Figure 1) may be identified with the notion of nominalisation (agentive in the local hub presented in the figure), while the leftmost block presents the single particle-attached affixation, as in one-upper, the middle one—the single verb-base-attached affix option, as in looker-on, and the rightmost one—double affixation, as in goer-awayer. The local hub is a small portion of a much wider and more complex network of -er suffixation possibilities in the English language, which will vary in terms of semantic relation (the various readings associated with this suffix) and in terms of the nature of the [[x] [y]] option, as, in many cases, there will be only an [x] (single base variable), as in teacher, or other elements may be added [[[x] [y] [z] -er], as in in-the-parker and up-and-comer), etc.

This allostructional family is available to speakers to choose from in any instance of particle verb nominalisation, and the analysis of the data (the types found in the three sources and the ones from the web as corpus) did not reveal any single factor as a candidate for an absolute or even strong constraint on the choice of double -er marking.

Denison (2008, p. 2014) hypothesised that -er doubling tries to avoid homonymy with a comparative form of an adjective (upper, inner, outer, and rounder), but the data analysed do not bear directly on such a hypothesis. First, downer is not a comparative form of an adjective, but a noun, inner, outer, offer, and upper are also nouns, and more importantly, as independent words, upper and downer have meanings fully compatible with the meaning of some nominalised particle verbs with a stimulus reading, e.g., picker-upper, bringer-upper, and calmer-downer. The existence of homonymous full words may be considered as a strengthening factor in -upper constructions (as they are the most numerous in all sources) and even suggest a compounding origin story. It is probably not by chance that OED (Oxford English Dictionary 2023) identifies -upper as “a combing form (i.e., a compound constituent), forming rhyming compounds based on phrasal verbs with up adv. with the agent noun corresponding to the verb as the first element, as fixer-upper n., maker-upper n., picker-upper n., etc”. Downer, inner, offer, and outer also exist as independent lexemes (admittedly, offer stands apart as the most homonymous, i.e., no semantic association can be established between the noun offer and the nominalised -off constituent), but none of them compare to -upper in terms of types of doubler-upper nominalisations (although nominalised -out boasts significantly more instances than any of the other analysed particles, with the exception of -upper). This should come as no surprise, since Dixon ([1991] 2005, p. 346) states that “up, out and off—these are in fact the most commonly occurring prepositions in phrasal verbs. Others are down, in and on, which are the next most common prepositional components”. This means that it is not homonymy with a comparative adjectival form that is the leading factor, but the productivity of the respective particles for constituting different types of particle verbs. The role of homonymy for doubler-uppers can be interpreted as one among multiple motivations for the constructions, rather than a dedicated homonymy-resolution strategy. In other words, neither a strengthening nor a constraining effect can be postulated across the board in relation to homonymy. The most frequent prepositions in particle verbs logically display the greatest number of types of doubler-upper nominalisations.

In both antonymous pairs up vs. down and in vs. out, any expectation of mutual coercion remains unsupported. Upper and downer nominalisations show as stark a contrast in terms of types as inner and outer do. Symmetry supported by particle antonymy does not seem to be an influential factor, especially considering that awayer construction types are not remarkably fewer than -inner ones.

Both breaker-upper (reported to have an idiomatic, i.e., figurative meaning), and goer-awayer (identified as a compositionally constructed particle verb with spatial meaning) exist (among numerous other types), which implies that figurativity in the semantic composition of the particle verb does not seem to be a facilitating or an adversely interfering factor determining the choice of suffixation alternative.

The distinction between prepositions and adverbs in the constituency of particle verbs also does not seem to have an immediate effect on the choice or not of double -er marking. Maker-outer is derived from a verb and an adverb particle and has figurative meaning in the sense from which the nominalisation is derived, while adder-toer is derived from a verb–preposition combination and is not characterised by idiomaticity or figurativity, and putter-upper-wither includes triple affixing (on each constituent of the MWV) and is figurative in meaning. It can be cautiously concluded that the number and status of the non-verb constituent(s) in a particle verb do not constrain or determine the possibility of double or multiple -er marking (depending on the nature of the base).

Transitivity also does not appear be an operationally constraining factor for doubler-uppers. Double -er marking attaches equally easily to both transitive and intransitive particle verbs: filler-outer_tr, eater-outer_intr; (gut) sucker-inner_tr, sleeper-inner_intr.; ticker-offer_tr., shower-offer_intr.; disher-upper_tr., messer-upper_intr.; taker-downer_tr., sitter-downer_intr.; and giver-awayer_tr., runner-awayer_intr., despite the existence of well-established verb-attached single suffix alternative nominalisations. Go out and go away are intransitive MWVs (unlike the transitive go for) and both goer out and goer away exist as alternatives to doubler-uppers. Blow up is a transitive MWV and both blower-up and blower-upper are actualised alternatives. The predominance of nominalisations from transitive MWVs, as can be evidenced in

Table 2. Percentage of nominalisations from transitive and intransitive MWVs within the dataset.

Particle	Percentage of Nominalisations from Transitive MWVs	Percentage of Nominalisations from Intransitive MWVs
-up	26.5%	73.5%
-down	20%	80%
-in	47.7%	52.3%
-out	26%	74%
-away	77.8%	22.2%
-off	83.3%	16.7%
Total	27.3%	72.7%

below, is exclusively dependent on the semantics of the combination of the verb base and particle in the MWV and does not seem to be correlated with a specific allostructional choice, which suggests that the said choice is not sensitive to transitivity.

It may also be mentioned that there is a tendency for recursive expansion of the MWEs via the incorporation of objects in certain nominalisations, such as football-snatcher-awayer, cab-runner-awayer, spoon-putter-outer, trash-taker-outer, and problem-hasher-outer, etc., which confirms the unified status of doubler-upper particle verb nominalisations as word elements, which is also supported by the tendency for hyphenation in their spelling.

It might be stipulated that there are two further motivating factors for doubling the suffix -er. One of them is the existence of a comparative correlative construction, which purely digrammatically might support doubling. The second is the plausibility in certain doubler-upper types (especially with -upper and -downer) of a compounding derivational interpretation. Even though Lensch (2018, p. 159, footnote 2) insists “that -er simultaneously attaches to verb and particle/preposition in the case of doubler-upper nouns” and that they do not represent “a case of compounding in which a nominalized verb combines with a nominalized particle/preposition”, it is not improbable that the latter might have been an initial motivating factor. These two-fold affixed nominalisations might have started as affixed verbs compounded with upper or downer, and from there, analogy has strengthened the case of doubling the marker across other particle verbs. It has long been recognised that there is a tendency for “[a] bipartite marker […] to replace an isofunctional marker consisting of only one of the two elements, i.e., the complex marker replaces the simple marker […]; my translation—SRA)” (Anderson 2015, p. 278, quoting one of Kuryłowicz’s laws of analogy). This plausible motivation story, not a diachronic origin story, is in keeping with the ideas of lexical relatedness and multiple motivations for lexical items via multiple existing schemas (Booij and Audring 2018).

Specific quantitative information in relation to any of the factors mentioned above would make the claims in this qualitative analysis sufficiently more robust and is the naturally occurring subsequent stage of this work in progress. The same applies to the patterns of polysemanticisation of the doubler-upper construction—agentive, patient, entity, and eventive readings (although the place and inhabitant readings typical of the suffix -er are definitely lacking in any of the analysed types) to add to the study of this polyfunctional suffix in English that can attach to “verbs and nouns, but also to adjectives, prepositions, verb-particle constructions (sometimes with an “extra” -er) with the particle preceding or following the verb and if following the verb, preceding or following the -er suffix, and a variety of phrases, including various kinds of noun phrases, prepositional phrases, adverb phrases, conjoined phrases, verb phrases and a handful of less easily categorized constructions” (Ryder 2000, p. 293).

4. Conclusions

Doubler-upper nominalisations appear to be deviant, both lexically and derivationally, and, at the same time, optimal in maintaining a balance between complexity and unity, which, in many cases, may also be supported by optimal rhythmic considerations. If we subscribe to the postulation that “[t]he form of a derivative is optimal if it is the best one morphophonologically” (Hathout and Namer 2014, p. 186), then we need to redefine ‘morphophonologically’ in terms of naturalness and semantic fitness. These paradoxical claims may both be separately and conjointly true, if we recognise the dynamicity of language as a complex, emergent, adaptive system. Such kinds of paradoxes and phenomena, akin to the flexibility of nominalisations of particle verbs in English, are captured by Libben’s (2021) interpretation of the lexicon as a flexicon comprising words which are themselves dynamic actions, where two principles operate to maximise naming possibilities and storage capabilities—morphological transcendence and superstates.

Double (or multiple) -er marking (where marking is intended to collapse any commitment to process associations—compounding, affixation, reduplication, or a mixture of these) on particle verbs in English results from the suffix -er developing lexical superstates via morphological transcendence, without leading to fixed morphological structures as Libben (2021) claims, since words are actions, not static representations stored in the mind. The notion of the flexicon captures both onomasiological and semasiological perspectives in word processing. In the speaker’s view, a superstate realisation is epiphenomenally shaped when searching for a way to encode a target meaning in a specific communicative situation, while the specific realisation in context triggers relations to a superstate in comprehension and models the relations between item-specific and schema-general knowledge. This dynamicity and flexibility can be analytically captured by applying an allostructional approach with a balanced “division of labour between sister links and mother schemas” (Audring 2019, p. 293).

The still-questionable status of doubler-uppers as ‘emancipated’ colloquialisms within a tripartite competition of free speaker’s choice points towards a very complicated relationship between entrenchment (the individual speaker’s perspective) and conventionalisation (the communal status and distribution of types and tokens of such constructions), which are interrelated via feedback loops into usualisation frequencies (see Schmid (2020) for elaborate analyses of these dimensions of dynamic language). The use of doubler-uppers has been claimed to be on the rise diachronically (Chapman 2008, pp. 272, 276, 280) and in terms of permeation in written genres, more specifically British and American newspapers, as claimed by Lensch (2022, p. 83), but considering that very little is predictable in language (Langacker 2008), not much can be said about their immanent conventionalisation.