Abstract
This paper reports a case study on a family of American English constructions that will be called the family of approximate comparison constructions. This family has three members, all of which follow the syntactic pattern about as X as Y with X being an adjective, but which allow three related functions: literal comparison, simile and irony. Two cognitive frameworks concern themselves with irony, the cognitive modelling approach and viewpoint approach, and the paper will show that, while the ironic approximate comparison construction calls central assumptions of the cognitive modelling approach to question, the viewpoint account can be refined to handle these cases. In doing so, it furthers our understanding of the cognitive underpinning of irony. The paper provides a corpus-based analysis on the Y slot as well as collostructional analyses on the adjectival X slot in the family of approximate comparison constructions. The results thereof suggest that the ironic approximate comparison construction, in comparison to its literal counterpart, prefers adjectives that convey positively connotated, nuanced attitudes and is formally less variable in the Y slot. The preference for particular adjectives lends further support to the assumption that hearers understand the construction as ironic or literal before speakers complete their utterance. Given that, it is argued that the ironic approximate comparison construction communicates an inherent viewpoint.
1 About as X as Y – a starter
The syntactic string about as X as Y can convey different, albeit related, meanings, as the following American English examples (1)–(3)[1] illustrate:
| He urged his tired horse in the direction of the lights, then picked up an eyebrow of trail that led to an undercut limestone shelf about as high as a tall pine. |
| Let’s begin though this morning with the growing controversy in the world of fashion. We all know that when it comes to modeling and advertising, air brushing is about as American as Apple Pie. |
| David was a lawyer, but he worked for a record company, on its business side. He was nice but about as interesting as iceberg lettuce. |
While example (1) illustrates a fully compositional comparison between a shelf and a pine in terms of height, examples (2) and (3) convey figurative comparisons, whose meanings cannot be derived on an item-by-item basis. Example (2) illustrates the use of about as X as Y as a simile, i.e., “a[n asymmetric] predicative comparison” (Bredin 1998: 75; see also Dancygier and Sweetser 2014: 141–142) in linguistic terminology. The coloring technique air brushing is predicated with the property of being about as American as Apple Pie. Example (3) illustrates the use of about as X as Y as an ironic simile. Irony always involves some kind of “mismatch” and the mismatch here is between entities that would “normally” be categorized as interesting and iceberg lettuce, which is a non-prototypical member of the aforementioned ad hoc category (see Veale and Hao 2010, who draw on the concept of ad hoc categories introduced by Barsalou [1983]). Given such a mismatch, the ironic meaning of about as interesting as iceberg lettuce is evident without the aid of further contextual clues.
The three uses of about as X as Y are best understood as a family of constructions (Goldberg 1995; and contributions to Ruiz de Mendoza et al. 2017), henceforth called the family of approximate comparison constructions. Constructions are defined in the Goldbergian sense as “conventionalized pairings of form and function” (Goldberg 2006: 3). Examples (1) to (3) share the same basic surface form, which provides one of the reasons for treating them as a family. However, they fulfill seemingly different functions.
The aim of the present paper is twofold. The first aim is to show that about as X as Y indeed forms a family of constructions related by subpart links. Using a corpus-based approach, it will be shown that the literal use of the approximate comparison construction is formally more variable in the Y slot, but, at the same time, also less productive than the ironic use as regards the Y slot. Furthermore, a simple collexeme analysis of the adjectival X slot is reported, which suggests that the literal approximate comparison construction prefers rather general adjectives and adjectives denoting spatial dimensions, while the ironic approximate comparison construction prefers adjectives that convey positive and nuanced assessments of a state-of-affairs. Furthermore, a distinctive collexeme analysis of the adjective is reported, which suggests that there are quite a few adjectives that are associated with the ironic approximate comparison construction. Given the fact that these adjectives have projecting power as to the overall interpretation of the construction, it will be argued that hearers can infer the ironic meaning of the utterance before its completion. The second aim is to draw conclusions from these results for a broader, cognitive account of irony. It might be levelled against this paper that it reports a case study on a non-prototypical form of irony and, thus, that its findings are non-generalizable. Still, given the fact that the existing accounts of irony have been developed on the basis of prototypical examples, considering non-prototypical cases represents a litmus test for their soundness. Two cognitive frameworks lend themselves, the cognitive modelling account of irony (Ruiz de Mendoza 2017; Ruiz de Mendoza and Lozano-Palacio 2019a, 2019b; Ruiz de Mendoza and Masegosa 2014) and the viewpoint account (Dancygier 2017; Tobin and Israel 2012). In the light of the results gained from the constructional analysis, it will be argued that the former cannot handle all cases of the ironic approximate comparison construction, while the latter can be refined in such a way that it can explain the observations made.
Accordingly, the paper is structured as follows. Section 2 will briefly introduce and review the cognitive modelling and the viewpoint account of irony. In Section 3, the corpus approach the paper uses is explained in more detail, including a brief introduction to collostructional analyses. Section 4 reports on the results of these corpus analyses and, in doing so, provides a detailed constructional analysis of the family of approximate comparison constructions. Based on these findings, conclusions on how hearers and readers interpret about as X as Y on-line will be drawn. Section 5 will address the question whether the cognitive approaches to irony outlined in Section 2 can be squared with the insights gained in Section 4. It will be argued that, while some aspects of the cognitive modelling approach can hardly be reconciled with the data, the viewpoint account of irony is compatible with them. Finally, the paper will be summarized in Section 6.
2 Two cognitive approaches to irony
Research on verbal irony has a long tradition in pragmatics, starting with some remarks in Grice’s (1989) seminal paper. It lies outside the scope of the present paper to provide a comprehensive overview on these (but see Garmendia 2018). In cognitive linguistics, two major strands of research concern themselves with verbal irony. One is the cognitive modelling approach (Ruiz de Mendoza 2017; Ruiz de Mendoza and Lozano-Palacio 2019a, 2019b; Ruiz de Mendoza and Masegosa 2014), the other is the viewpoint approach (Dancygier 2017; Tobin and Israel 2012). The following section will review both.
2.1 The cognitive modelling approach to irony
The cognitive modelling approach to irony (Ruiz de Mendoza 2017; Ruiz de Mendoza and Lozano-Palacio 2019a, 2019b; Ruiz de Mendoza and Masegosa 2014), builds on the notion of idealized cognitive models, i.e., abstract conceptual structures based on experiences of reality (Lakoff 1987). One particular kind of idealized cognitive model that is relevant for irony is the propositional model, which, in its simple form, contains information on one or more elements and their relations. Propositional models have otherwise become known as ‘frames’ (Cienki 2010; Fillmore 1982; Ruiz de Mendoza and Masegosa 2014). More complex propositional models can be structured by scenarios, which may specify several simple propositional models and the inherent order in which these have to follow.
The cognitive modelling approach to irony makes use of the notion of scenarios. It assumes that irony involves the activation of (at least) two scenarios, an observed and an echoed one. In doing so, it makes explicit use of Wilson and Sperber’s echoic allusion account of irony (Wilson and Sperber 2012: Ch. 6) and regards itself a cognitive extension thereof, albeit with some modifications. It maintains that echoing (rather than alluding) is the more adequate description of what the ironic speaker does since allusions require an antecedent while echoes do not. A cognitive echo, then, is defined as “the repetition of a thought (i.e., an internally consistent and meaningful conceptual complex) designating a real or constructed state of affairs” (Ruiz de Mendoza 2017: 189) and this thought is in some way at odds with what is observed. When the two scenarios clash, assumptions that are part of the echoed scenario are cancelled. This cancelling process also explains the attitudinal factor observed for irony since the speaker revises prior beliefs about a state-of-affairs, and in doing so, their attitude towards this process becomes apparent, e.g., in the form of disappointment, frustration, impatience, and the like.
As becomes obvious, one central, if not the most important, component of the cognitive modelling account of irony is the echoed scenario. At a first glance, it is hard to find the thought that might be repeated in example (3) He was nice, but about as interesting as iceberg lettuce. Ruiz de Mendoza (2017) treats constructs of the approximate comparison construction as special cases of echoing and cites the simile about as useful as a microscope and the ironic about as useful as buying one shoe. He argues that the simile echoes a stereotypical belief, i.e., that microscopes are useful, while the ironic construct echoes its opposite. Likewise, example (3) could be analyzed in such a way that the adjective interesting sets up an expectation that a salient representative of the category ‘interesting’ is following, which can be treated as the repeated thought. In other words, the male referent is, initially, ascribed this property. This expectation then clashes with what is actually uttered and triggers the ironic effect. As a part of this effect, the initial assumption of the echoed scenario – that the male referent is someone interesting – is cancelled and replaced by a revised assumption, i.e., that he is not interesting at all. The hyperbole created by echoing the opposite of the stereotype, iceberg lettuce, reinforces the negative assessment made about the male referent and makes the speaker’s attitude even more apparent (see Popa-Wyatt 2020 on the relationship between irony and hyperbole).
2.2 The viewpoint account of irony
The viewpoint account of irony (Dancygier 2017; Tobin and Israel 2012) builds on the notion of mental spaces (Fauconnier 1994, Fauconnier 2010) and conceptual integration (Fauconnier and Turner 2002; Turner 2010, 2018). Mental spaces are cognitive meaning packages set up for local understanding as discourse unfolds. Originally, they were introduced to explain phenomena such as anaphora or counterfactuals and their resolution in discourse: Uttering If I were President of the United States, for instance, sets up two mental spaces. In one space, which might be dubbed the base space, the speaker is different from the President of the United States, while in the other space, the counterfactual space, the two entities are constructed as the same person. Crucially, the entities in the counterfactual space are mapped with their counterparts in the base space. The remote conditional construction serves as a so-called ‘space builder’ since it triggers the set-up of the counterfactual space. More often than not, mental spaces are structured by frames, scenarios or other kinds of cognitive models. In the example, If I were President of the United States, the idealized cognitive model of President of the United States, which includes the knowledge about head of states, executive power, democracy, etc., is essential in understanding. Conceptual integration theory is an extension of mental spaces theory to account for creativity. It assumes that speakers make creative use of at least two input mental spaces and selectively project elements and relations from these spaces to a novel space, the so-called ‘blended’ space, which follows its own emergent logic.
Irony is a case in point. An ironic utterance blends at least two input spaces. One of these is the base space (which may be the reality space or one that is construed as such) and the other one is some kind of alternative space (see also Coulson 2005; Pálinkás 2014). These two spaces each contain an element or a relation that is in some way contrary to its counterpart. When they are blended, the ironic utterance alludes to this distinct element and triggers a viewpoint shift (see Dancygier 2017; Tobin and Israel 2012). Example (3), he was nice, but about as interesting as iceberg lettuce, blends the base space that works as the reality space in the novel and an alternative space. The base space includes at least the male referent, frame semantic knowledge about iceberg lettuce, and frame semantic knowledge about the property ‘interesting’. The alternative space includes the same basic elements, but the frame semantic knowledge about iceberg lettuce is manipulated in such a way that it is construed as an interesting entity. When the spaces are blended, the mismatch between base and alternative space in their treatment of iceberg lettuce becomes obvious and a shift in viewpoint to a higher level is triggered. From this higher-level perspective, the inference can be drawn that the male referent cannot be genuinely described as someone interesting by mapping the mismatch back to the base input space.
The blending approach to irony has been criticized for two main reasons. Ruiz de Mendoza and Lozano-Palacio (2019a, 2019b), with particular reference to the mental space approach by Coulson (2005), argue that it neither acknowledges the existence of an observable scenario nor does it explain the attitude the ironic utterance conveys. This criticism cannot be levelled against later approaches, including in particular the one proposed by Tobin and Israel (2012). As explained above, their proposal includes a base space, in which an observable scenario takes place. This scenario is accessed from two different viewpoints: once from the lower-level viewpoint and once from a higher-level viewpoint. On the lower level, the male referent is construed as an interesting person, just like iceberg lettuce has an interesting, exotic taste. This lower-level viewpoint is possible since the blended space follows its own emergent logic. The frame semantic knowledge that iceberg lettuce is rather tasteless triggers the perspectival shift to a higher viewpoint space. On this higher level, the male referent is assessed as uninteresting, just like iceberg lettuce has an uninteresting, watery taste. This perspectival shift also explains the aforementioned attitude, because it implies an “alienation from this lower-level viewpoint and from those who hold it” (Tobin and Israel 2012: 34). Thus, conceptual integration can, in fact, explain the negative attitude observed for irony.
Both cognitive approaches to irony show a few remarkable parallels.[2] Both involve the construction of at least two conceptual packages, i.e., an observed scenario or a base space and an echoed scenario or alternative space, and a cognitive process triggered by such a construction, i.e., cancelling assumptions of the echoed scenario or a perspectival shift from a lower to a higher viewpoint. It seems, then, that the difference is a matter of conceptual hierarchy: Mental spaces and conceptual integration strive for explaining local creativity in discourse, while cognitive modelling elaborates on a more general scheme of irony based on local experiences. Their main difference, though, seems to lie in the cognitive processes they assume irony to involve, since cancelling assumptions and shifting the point-of-view are quite distinct mechanisms. The following corpus study on the family of approximate comparison construction will show that both approaches to irony, as they stand, cannot explain its ironic use in sufficient detail. It will be argued, though, that the viewpoint account can easily be refined in such a way that it can handle cases like these, while the cognitive modelling account of irony is less adept in its present form.
3 Method
The Corpus of Contemporary American English (COCA, Davies 2008–) was used to search for the syntactic string about as ADJ as, yielding 2,010 hits in total. Of these, 32 duplicates had to be discarded manually. The remaining 1,978 hits were categorized as literal comparisons, similes or ironies (=interpretation) using the working definition presented in Section 1. There were 1,175 literal comparisons, 163 similes, and 640 ironic uses.
As for the adjectival X slot, a simple collexeme analysis (Gries and Stefanowitsch 2004; Stefanowitsch 2013) was performed for all three members of the approximate comparison constructional family using the R package Collostructions (Flach 2017) to identify adjectives that are attracted or repelled by the construction in question by comparing the overall frequency of these adjectives in the reference corpus (here: COCA) with their occurrence in the construction. In addition, a distinctive collexeme analysis (Gries and Stefanowitsch 2004; Stefanowitsch 2013) was also performed. Distinctive collexeme analyses compare the frequency of expressions in one construction with the frequency of the same expression in a competing construction. In doing so, preferences for adjectives within the family can be identified.
In addition to the adjectival X slot, the Y slot was annotated for its syntactic form using The Cambridge Grammar of the English Language (Huddleston and Pullum 2002) as a reference. Possible values were clauses (C), verb phrases (VP), simple and complex noun phrases (simple NP and complex NP, respectively), adjective phrases (AdjP), adverb phrases (AdvP) and prepositional phrases (PrepP). A noun phrase was considered simple when it consisted of either a head only, a determiner plus head or a noun compound as head (with or without a determiner). All other noun phrases were considered complex.
4 The family of approximate comparison constructions
The following section is concerned with a constructional analysis of the family of approximate comparison constructions, using the results of the corpus-based analysis as evidence. At its core, Construction Grammar (Goldberg 1995, 2006; Hoffmann and Trousdale 2013) assumes that any linguistic form is inseparable from its meaning, and so it lends itself as a framework for analyzing the intimate relations between the different forms and functions of about as X as Y illustrated in Section 1. What is more, it will be shown that the different uses of about as X as Y form a family of constructions, i.e., a network of closely (formally and functionally) related constructions. The corpus analyses reveal that the family members show formal variations in the Y slot and have different preferences for the adjective they select in the X slot, which speaks to the assumption that the members are related by subpart links. As a consequence, some adjectives have projective power to the extent that hearers and readers of about as X as Y can infer either the literal or the ironic meaning of the syntactic string before its completion.
4.1 Commonalities
Formally, all the constructions share the syntactic template about as X as Y. As such, they are instantiations of the higher-level as X as Y construction, which is used to establish a comparison between two entities. One of these entities is either encoded in the subject position, when as X as Y occupies the complement position of the predicative construction, or is recoverable from context. Many labels have been proposed for this entity: ‘tenor’ (Fishelov 1993), ‘topic’ (Veale 2013) and ‘target (frame)’ (Dancygier and Sweetser 2014; Israel et al. 2004) of comparison. The other entity is encoded in the Y slot and functions as the ‘vehicle’ (Fishelov 1993; Veale 2013) or ‘source (frame)’ of comparison (Dancygier and Sweetser 2014; Israel et al. 2004). The property, on whose grounds the comparison is made, is encoded in the X slot of as X as Y and is usually called the ‘ground’ of comparison. In the present paper, the terms ‘source frame’ and ‘target frame’ will be used to refer to the comparands, in line with Dancygier and Sweetser (2014), to highlight the fact that the construction as X as Y profiles the mapping of a property from one entity (the source) to the other (the target).[3] These properties are inherited by the daughter construction about as X as Y.
The daughter construction about as X as Y crucially features the attenuator about. Moon (2008) claims that the daughter construction with about is always used ironically, while Veale (2013) maintains that it rather signals the speaker’s intention to be humorously creative in most cases, but may as well be a marker of mere imprecision. With the data at hand, the strong claim Moon (2008) makes cannot be supported. Example (1) above shows that about as X as Y allows being used as a literal comparison. What is more, such a use is quite frequent: As already described in Section 2, 1,175 of the 1,978 hits were considered literal comparisons and, thus, represent the largest group. Given that, Veale’s proposal to regard about as a marker of imprecision rather than a marker of irony is supported. About, then, signals that source and target frame bear a close, but not a striking, resemblance with regard to the ground.
4.2 Formal and functional variation
Despite the commonalities described in the previous section, it will be shown in the following that about as X as Y itself is but a family of constructions, called the approximate comparison construction here, with each member showing formal idiosyncrasies and collostructional preferences. Thus, the members are related by subpart links.
4.2.1 Symmetry between source and target frame
The first formal aspect of the family of the approximate comparison construction that will be discussed here is the (a)symmetry between source and target frame. Works on simile in general observe that the relation between source and target frame in literal comparisons is a symmetric one while in similes the relation is inherently asymmetric (Bredin 1998; Ortony 1979; Romano 2017). This applies to the different uses of the approximate comparison construction, too. In example (1) above, an undercut limestone shelf about as high as a tall pine, a limestone shelf (the target frame) is equated with a tall pine (the source frame) in terms of height. When the context of the utterance is considered, source and target frame are not interchangeable: the target of the comparison is necessarily the limestone shelf as it is the topic of the ongoing discourse in the context of the novel from which the example originates and is to be described in more detail. The pine, in contrast, is just introduced in the comparison construction itself. Even though it is not physically present in the given context, it still works as a source frame, because most language users have sufficient knowledge about the usual height of pines. It is, however, easy to construct a context in which the roles are reversed. If two foresters, for example, talk about the height of a pine that is absent in their current conversational context, one of the two might say The pine is about as high as the limestone shelf over there. In this example the shelf needs to be present in the physical context of the utterance, due to the fact that shelves, in contrast to pines, may vary considerably in their vertical dimensions and do not naturally lend themselves for height comparisons. What is important, though, is the fact that there is no intrinsic asymmetry in such a comparison. It is rather a matter of information structure (see also Dancygier and Sweetser 2014: 136, on (a)symmetries in copula constructions).
The use of about as X as Y as a simile and the ironic use, on the other hand, feature such an asymmetric relationship between source and target frame. In the case of simile, the source is a particularly salient, if not the most prominent representative of the property that is highlighted (see also Veale and Hao 2010). Using such a salient entity creates a vivid image, which is crucial for a figurative interpretation. In example (2), air brushing is about as American as Apple Pie, Apple Pie is used as a salient feature of being American, presumably because it is such a popular dessert. When this is transferred to air brushing, a remarkably vivid image is created. The kind of mapping found in similes is indeed asymmetric. Provided that the hearer of this utterance has a prior understanding of what ‘American’ means, including the knowledge that Apple Pie is a popular dessert, they will have no difficulties in creating the simile. The construction mirrors this asymmetry. Thus, it is possible to say that airbrushing is about as American as Apple Pie but not that Apple Pie is about as American as airbrushing.
The ironic use is likewise asymmetric, but for the opposite reason. In similes, the source frame is a highly salient entity while in the ironic use, the source frame is often incongruent with the frame encoded in the ground. Example (3), he was nice, but about as interesting as iceberg lettuce, illustrates this. Here, frame semantic knowledge about iceberg lettuce and about the property ‘interesting’ are structurally presented as compatible, but in fact they are not, because iceberg lettuce is no salient entity of the interesting-frame, and thus it is an incongruent choice. Iceberg lettuce is about as interesting as he is, therefore, only possible in contexts, in which the male referent has already been established as someone extremely uninteresting. This is also the difference to example (1), where role reversal is possible, too. In example (1), no mutually shared prior knowledge about the limestone shelf is necessary to understand the comparison. Reversing the roles of example (3), though, requires such knowledge.
Salience is a gradable notion, though, and so it is unsurprising that disputable cases can be found in the corpus, such as the following:
| In the morning, the doctors released Eve but cautioned us to keep alert for any odd behaviors, slurred speech, and the like. Of course, I’d already planned to keep a remote, although close, eye on her. The human brain is about as tough as a ripe avocado and the long-term effects of concussion are unpredictable. |
In this case, the adjective tough is used as the dimensionally positive polarity pole of the local opposites tough-soft.[4] As a consequence, the annotation of this particular example as a literal comparison, simile, or irony depends largely on the salience of avocados as a tough (or rather soft) entity. Avocados are arguably not highly representative for exemplifying softness, as can be seen from the fact that the noun needs to be modified by ripe. What is more, a search for the phrase soft NOUN in COCA returns only four hits for avocado, while other nouns, which, technically, could also be used to describe human brains in this respect, are more frequent, including dough (105), polenta (45), and butter (36). On the other hand, even though four hits in COCA for soft avocado seems rather low, it still speaks to the fact that it is not incongruent per se – if it were so, there shouldn’t be any occurrences at all. This rules out the possibility of example (4) being ironic. Accordingly, example (4) was annotated as a literal comparison.
4.2.2 Formal variation of the source frame
As described in Section 2, the Y slots of all occurrences of about as X as Y in COCA were annotated for syntactic form. The results are summarized in Table 1.
Syntactic forms of the source frame grouped by interpretation.
| AdjP | AdvP | Clause | Complex NP | Simple NP | VP | PrepP | Narrative | Total | |
|---|---|---|---|---|---|---|---|---|---|
| Comparison | 11 | 4 | 813 | 171 | 158 | 14 | 4 | 0 | 1,175 |
| Irony | 0 | 0 | 53 | 265 | 197 | 124 | 0 | 1 | 640 |
| Simile | 0 | 0 | 28 | 59 | 48 | 28 | 0 | 0 | 163 |
| Total | 11 | 4 | 894 | 495 | 403 | 166 | 4 | 1 | 1,978 |
Table 1 shows that the source frame of about as X as Y, when used to convey a literal comparison, can have various syntactic forms, including adjective phrases, adverb phrases, clauses, simple and complex noun phrases and prepositional phrases. What is striking is that, in the majority of cases (i.e., 69%), the source frame has the form of a clause. However, even though the token frequency of the clausal form is quite high, its type frequency is rather low. A considerable number of 619 of these clauses do not encode a particular entity, but allude to some kind of generic, often maximum expectation about the target frame:
| As you pull into this station, the high-rises of the Loop glimmer grandly, about three miles away. But directly below you are stripped cars, crumbling buildings and knots of haggard men camped in doorways. The contrast is about as stark as urban contrasts get. |
In example (5), the source frame is a rather abstract frame that encodes what can maximally be expected of an urban contrast (i.e., a commercial business district opposed to a run-down area). In other words, there is no second entity to which the target frame is compared, but the source frame is the same as the target frame, albeit on a higher level of abstraction. Other clausal forms that were found to fulfill such a function include it gets (127), it can/could get (34) and they come (31) and variations thereof. Given their high frequency, it can be assumed that these are instantiations of a lower-level construction related to the literal approximate comparison construction by instance links. Likewise, twelve of the fifteen adjective and adverb phrases found encode a source frame that is an abstraction of the target frame, the types being possible, ever, usual, and normal. When the source frames have the form of either simple or complex noun phrases, verb phrases or prepositional phrases, these are more productive.
When about as X as Y is used as irony or as a simile, the source frame is formally less diverse. In the present corpus study, no adjective phrases, adverb phrases or prepositional phrases were found. Also, compared with literal comparisons, less clausal forms were found. Instead, these two constructions seem to prefer source frames encoded as (simple and complex) noun phrases, or verb phrases, often with the verb in the gerund form:
| What he’d said about robbers and murderers made a long walk in these parts about as attractive as getting a root canal without Novocain. |
It seems, then, that the source frame of about as X as Y, when used to convey similes or irony, must be different from the target frame. There is only one exception from this general rule in the data:
| I own a crayon drawing Alexander made at about this age. His mother gave it to me – no big sacrifice on her part, since she saved all his drawings, and had hundreds of them. It’s about as impressive as you’d expect from a drawing made by a six-year-old: stick figures, lollipop trees, lopsided houses, garish color that refuses to accept the authority of the lines meant to hold it in check. |
In example (7), the source frame is the same as the target frame, but on a higher level of abstraction, and explicitly so (as you’d expect). What renders this example ironic, then, is the rather curious choice of the adjective impressive, which is hyperbolically at odds with what would normally be expected about a comment on a children’s drawing. Section 4.2.3 will elucidate the role of the adjective in more detail. Despite this exception, the general observation that source and target frame must be different entities is fully in line with previous research, since it is the unexpected connection made between two concepts which makes a comparison figurative (Israel et al. 2004). Therefore, the Y slot in about as X as Y, which encodes the source frame, requires forms that are, prototypically, used for introducing new entities and adjective, adverb, and prepositional phrases do not prototypically do so.
4.2.3 Collostructional preferences
The simple collexeme analysis for all three uses of the approximate comparison construction resulted in only one adjective being repelled by all of them, namely other. This adjective can be discarded from further analyses for reasons of blocking: in the construction, other would have the meaning ‘different’ and is probably blocked in favor of the adjective different. The following discussion will thus concentrate on attracted adjectives. Moreover, the simple collexeme analysis for similes proved to be rather unrevealing. Only four adjectives reached a significance level below p = 0.00001, which are close, passionless, common and thick. With such a low number, it is impossible to figure out commonalities. Moreover, due to its low constructional frequency (N = 163), the expected frequencies for single adjectives mainly range between 1 and 0 and thus no reliable conclusions can be drawn. For these reasons, similes have been excluded from further considerations.
The result of the simple collexeme analysis for the literal comparative use is summarized in Table 2.
Extract of the simple collexeme analysis performed for the literal approximate comparison construction, showing adjectives that are significant at p < 0.00001 (*****).
| ADJ | CORPUS FREQ | O | E | ASSOCIATION | COLL.STR.LOGL | STR.DIR | p |
|---|---|---|---|---|---|---|---|
| Close | 47,265 | 82 | 1.8 | Attr | 475.21479 | 475.21479 | ***** |
| Good | 388,225 | 115 | 14.5 | Attr | 284.83521 | 284.83521 | ***** |
| Big | 196,335 | 75 | 7.3 | Attr | 217.81955 | 217.81955 | ***** |
| Bad | 93,165 | 53 | 3.5 | Attr | 192.04707 | 192.04707 | ***** |
| Low | 65,790 | 30 | 2.5 | Attr | 95.85099 | 95.85099 | ***** |
| Wide | 35,928 | 22 | 1.3 | Attr | 82.23423 | 82.23423 | ***** |
| High | 191,989 | 42 | 7.2 | Attr | 80.06376 | 80.06376 | ***** |
| Thick | 21,661 | 18 | 0.8 | Attr | 77.65553 | 77.65553 | ***** |
| Fera | 105 | 4 | 0 | Attr | 47.61598 | 47.61598 | ***** |
| Tall | 22,748 | 13 | 0.8 | Attr | 46.81967 | 46.81967 | ***** |
| Bright | 27,225 | 13 | 1 | Attr | 42.47747 | 42.47747 | ***** |
| Tough | 35,096 | 11 | 1.3 | Attr | 27.55336 | 27.55336 | ***** |
| Large | 119,420 | 19 | 4.4 | Attr | 26.24913 | 26.24913 | ***** |
| Simple | 50,227 | 12 | 1.9 | Attr | 24.43227 | 24.43227 | ***** |
-
a Fer here has the same meaning as the adjective far in these contexts. In three of the four occurrences, it is part of quoted material and could be treated as eye dialect. In the fourth one, it could be a typo.
Table 2 only shows adjectives that are significant at p < 0.00001 (*****) for reasons of space. The two adjectives that occur only once in both the construction and the reference corpus (i.e., nonapologetic and unferocious) have been discarded, too. The analysis reveals that there are two major kinds of adjectives that are attracted by the construction. On the one hand, adjectives that denote some kind of relation appear more often than expected. These include relations denoting broad evaluations (good, bad), complexity (tough, simple) and mental or emotional proximity (close). On the other hand, adjectives that denote spatial dimensions are attracted, too, i.e. close (in the sense of ‘spatial proximity’) big, low, wide, high, thick, fer, tall and large. The only adjective that is not covered by these two categories is bright. Bright occurs 13 times in the literal use of the approximate comparison construction, and 12 of these feature its perceptional sense as in the following example:
| The Great Comet of 1901 reached prominence in April and remained visible to the naked eye for six weeks. It peaked at around magnitude -2, about as bright as the planet Jupiter typically appears, on April 24. |
With only two exceptions, examples like (8) occur in magazines addressing an audience interested in astronomy. Since presumably most members of the target audience are laypersons, technical information on magnitudes may not be useful and so the literal approximate comparative construction is used to illustrate the brightness of interstellar objects.
Taken together, the kinds of adjectives attracted by the literal sense of the approximate comparison construction suggest that, more often than not, structurally rather simple concepts are literally compared. This is certainly true for those adjectives that denote spatial dimensions: the size of an object is easy to grasp for most people and in case it is not, a literal comparison is quite useful (see example (1), for instance, where the size of the object could not be perceived directly since it was part of a novel). Bright, in the visual sense, is also a rather easy concept. Relations, in particular evaluative relations, like good and bad, are more complex than concepts that involve spatial dimensions or perceptions, but in the literal use of the approximate comparison construction they are constructed as simple:
| Mothers kill their children when they’re drunk. Or drink when they’re pregnant, which is about as bad as killing them. |
Individual value systems may diverge in many respects, but killing a child is generally seen as a reprehensible act, and so is drinking when being pregnant. Even though moral values can be complex systems, they are constructed as simple in this example.
The results of the simple collexeme analysis for the ironic use of the approximate comparison construction can be found in Table 3.
Extract of the simple collexeme analysis performed for the ironic approximate comparison construction.
| ADJ | CORPUS FREQ | O | E | ASSOCIATION | COLL.STR. LOGL | STR.DIR | p |
|---|---|---|---|---|---|---|---|
| Likely | 68,879 | 39 | 1.4 | Attracted | 186.69494 | 186.69494 | ***** |
| Exciting | 11,469 | 24 | 0.2 | Attracted | 175.91753 | 175.91753 | ***** |
| Useful | 19,785 | 18 | 0.4 | Attracted | 102.20167 | 102.20167 | ***** |
| Subtle | 8,646 | 14 | 0.2 | Attracted | 95.28912 | 95.28912 | ***** |
| Appealing | 2,910 | 11 | 0.1 | Attracted | 93.33886 | 93.33886 | ***** |
| Popular | 45,653 | 16 | 0.9 | Attracted | 61.37438 | 61.37438 | ***** |
| Threatening | 1,263 | 6 | 0 | Attracted | 53.60021 | 53.60021 | ***** |
| Sexy | 6,480 | 8 | 0.1 | Attracted | 50.09539 | 50.09539 | ***** |
| Accurate | 12,819 | 9 | 0.3 | Attracted | 46.42805 | 46.42805 | ***** |
| Easy | 58,801 | 14 | 1.2 | Attracted | 43.58581 | 43.58581 | ***** |
| Pleasant | 7,840 | 7 | 0.2 | Attracted | 39.37081 | 39.37081 | ***** |
| Welcome | 8,589 | 7 | 0.2 | Attracted | 38.12294 | 38.12294 | ***** |
| Reliable | 9,104 | 7 | 0.2 | Attracted | 37.32806 | 37.32806 | ***** |
| Effective | 39,290 | 10 | 0.8 | Attracted | 32.30541 | 32.30541 | ***** |
| Convincing | 3,656 | 5 | 0.1 | Attracted | 32.29494 | 32.29494 | ***** |
| Forgiving | 1,386 | 4 | 0 | Attracted | 31.74825 | 31.74825 | ***** |
| Interesting | 41,458 | 10 | 0.8 | Attracted | 31.31782 | 31.31782 | ***** |
| Inviting | 1,707 | 4 | 0 | Attracted | 30.09251 | 30.09251 | ***** |
| Predictable | 4,104 | 4 | 0.1 | Attracted | 23.16595 | 23.16595 | ***** |
| Edgy | 1,227 | 3 | 0 | Attracted | 22.81779 | 22.81779 | ***** |
| Maneuverable | 157 | 2 | 0 | Attracted | 21.80517 | 21.80517 | ***** |
| Authentic | 5,471 | 4 | 0.1 | Attracted | 20.92014 | 20.92014 | ***** |
| Appetizing | 225 | 2 | 0 | Attracted | 20.36074 | 20.36074 | ***** |
The analysis shows that the ironic approximate comparison construction attracts 22 adjectives most significantly. If the adjectives of this list are compared with those for literal comparisons, two main aspects become obvious. The first observation is that the adjectives used in the ironic approximate comparison construction do not denote any spatial dimensions or other kinds of objective or easily measurable concepts. Exciting, for example, denotes a property that is highly subjective: what is exciting for one person may not be as exciting for others. This is also what makes such properties more abstract than properties like spatial dimensions. Spatial dimensions can be measured easily, while properties like exciting cannot, due to their subjective tinge. This does not mean that spatial dimensions cannot be ironicized, as the following example illustrates:
| The mortality rate is high. That’s the way of the world when you are an inch long and about as fat as dental floss. |
Example (10) is about pipefish and the author humorously alludes to the fact that small animals like these rapidly fall victim to deadly causes. So here, an adjective denoting a spatial dimension (fat) is used to ironicize this state-of-affairs. However, the simple collexeme analysis suggests that this is unlikely. The ironic approximate comparison construction seems to attract adjectives that target subjective experiences. Research on irony suggests that irony is used to criticize an entity (Dynel 2013; Garmendia 2010, 2011, 2014). The act of criticizing involves a more or less subjective viewpoint, and so it is not surprising that the ironic approximate comparison construction attracts adjectives that already express this subjective element.
What is also striking in this respect is the observation that all but one adjective in Table 3 either express a positive (e.g., exciting, useful, popular, pleasant, welcome) or a neutral property (e.g. likely, maneuverable), while there is only one adjective that expresses a negative property (threatening). At first sight, this may lend support to the idea that irony can also be used to convey a positive assessment. If the contexts of the occurrences of threatening in the ironic approximate comparison construction are considered, this initial impression must be rejected, as the following example illustrates:
| Giving him a glare she hoped looked evil, she asked, “Why are you here?” His bemused smile said her evil glare was about as threatening as a cream puff. |
One might argue that the ironic approximate comparison construction is used to praise the female referent’s innocent look, but such an interpretation falls short of the more complex inferences that the construction generates in this context. The context maintains that the female referent’s intention was to look evil and threatening. For some reasons, though, she fails to do so and is subsequently humorously criticized for this failure. These observations also extend to the other instances of threatening, which are all used to criticize a state-of-affairs.
The second observation is that the adjectives preferred by the ironic use often express evaluative properties (like exciting, useful, popular, etc.), but the two general adjectives good and bad, which are attested for the literal comparisons, are not associated with the ironic construction (both adjectives are non-significant). One reason for this might be that good and bad are too general an evaluation to voice a nuanced criticism. In the ironic approximate comparison construction, it is the adjective that carries the interpretative load and serves as the anchor to search for possible inferences. General adjectives like good and bad are already quite vague and when these are used ironically, the number of possible inferences hearers may draw rises exponentially. Therefore, it seems plausible that the ironic construction prefers adjectives with a smaller scope.
The distinctive collexeme analysis that was performed on the literal and the ironic approximate comparison construction is summarized in Table 4 and provides corroborative evidence.
Extract of the distinctive collexeme analysis.
| ADJ | LITERAL | IRONIC | ASSOCIATION | COLL.STR.LOGL | p | SHARED | ||
|---|---|---|---|---|---|---|---|---|
| O | E | O | E | |||||
| Big | 75 | 49.2 | 1 | 26.8 | Literal | 58.35647 | ***** | Y |
| Good | 115 | 80.3 | 9 | 43.7 | Literal | 57.37008 | ***** | Y |
| Bad | 53 | 34.3 | 0 | 18.7 | Literal | 46.97563 | ***** | N |
| Close | 82 | 57.6 | 7 | 31.4 | Literal | 38.39995 | ***** | Y |
| Low | 30 | 19.4 | 0 | 10.6 | Literal | 26.37339 | ***** | N |
| ADJ | LITERAL | IRONIC | ASSOCIATION | COLL.STR.LOGL | p | SHARED | ||
| O | E | O | E | |||||
|
|
||||||||
| Appealing | 0 | 7.1 | 11 | 3.9 | Ironic | 23.05012 | ***** | N |
| Popular | 1 | 11 | 16 | 6 | Ironic | 26.85634 | ***** | Y |
| Useful | 0 | 11.7 | 18 | 6.3 | Ironic | 3.78491 | ***** | N |
| Exciting | 2 | 16.8 | 24 | 9.2 | Ironic | 38.20194 | ***** | Y |
| Likely | 4 | 27.8 | 39 | 15.2 | Ironic | 59.56819 | ***** | Y |
Table 4 is divided into two parts. The upper half shows the adjectives that are associated with literal comparisons, while the lower half shows those that are associated with irony. The adjectives that are associated with literal comparisons confirm the observations made above. Literal comparisons seem to be associated with adjectives denoting spatial dimensions (big, low, and close) and adjectives denoting general evaluations (good and bad). Adjectives associated with ironic comparisons, though, are rather associated with adjectives denoting a more precise evaluation than the vague evaluations associated with literal comparisons. Moreover, ironic comparisons seem to target emotional matters (appealing, popular and exciting), the usefulness of objects or the probability of events.
4.3 On the interpretation of approximate comparison constructions
Section 4.2 has shown that the three uses of about as X as Y constitute a family of constructions related by subpart links. The present section deals with the question of how language users derive the meaning of these constructions, if they prefer one use over the other and under which circumstances they do so. It will be argued that hearers may already generate an ironic interpretation before the construction is encountered in its totality. The consequences this has on accounts of irony is subject of the subsequent Section 5.
Giora and colleagues have shown that constructions which attenuate highly positive concepts, such as It’s not the most exciting video ever trigger a sarcastic interpretation by default (Giora and Becker 2019; Giora et al. 2015a, 2015b, 2018). The ironic use of the approximate comparison construction might be one of these constructions as it is attenuated by about and, as suggested by the collostructional analyses presented in Section 4.2.3, prefers positive concepts in the adjectival slot. According to Giora and colleagues, a construction must be schematic rather than substantial, potentially ambiguous between its uses, and to be interpreted as such without contextual clues to count as default, though. The present corpus data does not support the view that the ironic approximate comparison construction may be preferred over the other uses in the interpretative process (see also Giora et al. 2018, interpreting the results presented in Veale [2013]). In the vast majority of cases, the ironic use features an “internal incongruity” (Partington 2011) and, thus, these cases are not ambiguous. There are some rare, but notable exceptions to this:
| I’ve been told to rejoice that my daughter went to heaven unmarked by sin, her soul clean and pure, perfect. That God has a special place reserved for her. I have been told that I should feel privileged: I have my very own baby angel, my own divine connection. To me, these are cartoonish images. They are about as comforting as imagining God as a robed elderly man with a long white beard, floating around on a cloud. |
In this example, no internal incongruity can be found as a divine image could, actually, be comforting to some people. However, its ironic interpretation is highly supported by its context, and so are all examples without internal incongruity in the present data. Here, the word cartoonish already conveys a negative evaluation, which stands in contrast to the positively connotated adjective comforting. In other words, the incongruity is not internal, but external to the construction in this example. As a consequence, it is still open to debate whether the ironic interpretation of the construct in (12) is triggered by the construction itself or by other factors.
While there is no support for assuming that approximate comparison constructions are interpreted as ironic by default, there is substantial evidence that its interpretation is projected in some cases and that hearers understand the construction as ironic before its completion. Auer (2009) argues that language comprehension is an inherently dynamic processes and elaborates on its temporal aspects. His arguments revolve around spoken language only, still, the notions central to the present discussion can be applied to the comprehension of written language, too. Auer shows that spoken language is synchronized, i.e., that as soon as the speaker starts talking, the hearer does not wait until the completion of the utterance, but starts making guesses about its possible meaning. The same applies to reading: Readers do not read until the end of a sentence to start the interpretative process, but make hypotheses while reading it. One outcome of this synchronization of (spoken) language is that both speakers and hearers make active use of projections (Auer 2009). Projecting elements are expressions that lead hearers (or readers) in conversations by creating syntactic and/or semantic expectations. The approximate comparison constructions contain such projective elements. On the one hand, the use of about as syntactically projects the occurrence of the adjective, of as, and a number of syntactic forms that may enter the Y slot, as shown in the previous subsections, since these are highly likely continuations.
What is more, though, certain adjectives can project the likely interpretation of the higher-level construction and, thus, help disambiguating the lower-level construction. The distinctive collexeme analysis reported in Section 4.2.3 shows that there are certain adjectives associated with either the literal approximate comparison construction or its ironic counterpart, and these adjectives project the construction’s meaning. If, for example bad is used, the likelihood of the construction to be literal is higher than 99%. Vice versa, if the adjective useful is used, the likelihood of the construction to be ironic also approximates 100%. In other words, when adjectives like these two are used in the adjectival X slot, the hearer (or reader) can already make strong assumptions about what will follow the adjective and does not necessarily rely on an incongruent element to derive the ironic meaning. In such cases, the Y slot is rather uninformative for generating the intended propositional meaning. Saying that something is about as useful as a fork in sugar bowl or about as useful as an appendix is certainly humorous (see Veale 2013), and, therefore, indispensable for irony, but neither provides processable information about how useful (or useless) something is from the speaker’s point of view.
As a result of projection, the perceived humorousness of the construct might be affected. Giora and Becker (2019) cite experimental evidence that default interpretations are experienced as less humorous than non-default interpretations. This might apply to well-established constructs of the ironic approximate comparison construction, too. Although the present data supports the assumption that the ironic construction is highly productive, there are a few constructs occurring more than once, one of these being about as interesting/exciting as watching paint dry (7). Given their higher level of entrenchment, it might well be that these are perceived as less humorous than other, more creative constructs. The same reasoning can be applied to constructs using strongly associated adjectives, irrespective of what follows in the Y slot. Due to their projecting nature, the hearer is alerted to the upcoming incongruent element and may find the construct less humorous than constructs using adjectives that are less strongly associated.
5 About as X as Y and cognitive approaches to irony
Accounts of irony in general and cognitive accounts in particular must be able to explain the findings on the family of approximate comparison constructions reported in Section 4. The present section discusses whether the cognitive modelling and the viewpoint approach can do so. As for the cognitive modelling approach, it will be argued that both the assumption of an echoed scenario and of that cancellation processes can be questioned. The viewpoint account, though it needs some refinement, can easily handle the ironic approximate comparison construction.
As outlined in Section 2.1, the cognitive modelling account of irony assumes that the ironic utterance evokes an echoed scenario, of which some assumptions will be cancelled while deriving the ironic meaning of the utterance. With regard to the approximate comparison construction, Ruiz de Mendoza (2017) argues that constructs, such as about as useful as buying one shoe, echo the opposite of the stereotype of what typically counts as useful. The collostructional analyses reported in Section 4.2.3, however, shows that some adjectives, including useful, are strongly associated with an ironic use. In fact, useful occurs exclusively in the ironic approximate comparison construction. Versatile language users have innate knowledge about such frequency distributions and often find examples that violate these frequency constraints curious, if not unacceptable (cf. statistical preemption, Boyd and Goldberg 2011; Goldberg 2011, 2019). Therefore, it might be rather unlikely that hearers activate the literal meaning of these cases. Giora and colleagues (Giora and Becker 2019; Giora et al. 2015a, 2015b, 2018) have shown that sarcastic constructions are interpreted as such by default, i.e., by bypassing the literal interpretation. Even though about as X as Y might not always be interpreted as ironic by default, it seems likely that the literal meaning is bypassed when using projecting adjectives like useful. And, crucially, without the literal meaning, there is no echoed scenario in these examples either. It rather seems that the intended meaning of the ironic approximate comparison construction is entrenched, given that the “right” adjectives are being used.
Even in cases in which an echoed scenario can be constructed, this echoed scenario need not be instrumental in inferring the intended meaning, nor do assumption need to be cancelled. Consider the following example:
| I am so close to being done with the Republcan [sic] party. They’re about as useful as a tick giving you a blood transfusion. |
The context shows unambiguously that the first-person author is not favorably inclined towards the Republican party. When experienced language users, who know that useful is used ironically in the approximate comparative construction, read it in such a context, they don’t need to assume that the author might, in fact, find the Republican party useful. At this point of the contribution, it is quite clear that useful means ‘useless’ from the author’s point of view and readers are aware of this. Thus, there is no echoed scenario necessary to infer the author’s intended meaning. What could be constructed, though, is the scenario that there are other people who find the Republican party useful. This is the repetition of a thought (attributed to the followers of the Republican party) designating a constructed state of affairs, and, thus, fits the definition of an echoed scenario. Despite the fact that this is an optional inference, its assumptions need not be cancelled. Rather, an additional inference is made, namely, that people holding such a belief are wrong from the author’s perspective. Thus, interpreting the ironic approximate comparison construction does not necessarily rely on constructing an echoed scenario nor on cancelling parts of this scenario.
The viewpoint account of irony can handle such cases. When the ironic use of the approximate comparison construction is encountered for the first time, two input spaces are blended. One of these spaces is set up by the adjective, which assigns a property to the target frame. The other space is set up by the construction’s Y slot and encodes the source frame, which turns out to be a non-prototypical representative of the property. This incongruence triggers a viewpoint shift from a lower to a higher viewpoint. While on the lower level, hearers follow along the lines of the literal ‘garden path’ and assign the property to the target frame, they are then inevitably prompted to reconsider the initial interpretation. They resort to a higher-level viewpoint and reinterpret the assessment in such a way that the target frame is assigned the opposite of that property (Colston 1999).
Accordingly, if the ironic approximate comparison construction, exemplified by (13) about as useful as a tick giving you a blood transfusion is encountered for the first time, two input spaces are set up. One input space includes the target frame ‘Republican party’ and this element is assigned the property of being useful. The other space is set up by the following Y slot and encodes the source frame ‘tick’, which is also assigned the property of being useful. On this viewpoint level, the hearer is misled into thinking that both the Republican party and ticks might genuinely be described as useful. However, this is in contrast to the idealized cognitive model most people have of ticks, and, therefore, a viewpoint shift to a higher level is initiated. From this higher-level viewpoint, the target frame, the Republican party, is reassessed as useless.
Although mental spaces and the blends they may create are local meaning constructs built on frames, they may become entrenched frames when encountered frequently (Fauconnier 2010). The corpus investigation presented in Section 4 suggests that this is the case for the ironic approximate comparison construction. Here, the blend that is created is a function connected to the linguistic form by convention (similar to the XYZ construction, Fear is the father of cruelty, Fauconnier and Turner 2002; Turner 1998). As a consequence of the blending template being entrenched, it seems that the input spaces get backgrounded. When the ironic construction is used with strongly associated adjectives, the blended structure gets activated, but not necessarily all component parts, such as the input space that is prompted by the adjective on first encounter. This is why useful in examples like (13) can be initially understood as useless. For obvious reasons, then, there is no viewpoint shift in these cases anymore, but the intended viewpoint is inherent to the construction and also activated immediately. This may also explain why default interpretations are perceived as less humorous. A shift in viewpoint is a cognitive affordance that leads to humor (Coulson 2015; Coulson et al. 2006) and when the viewpoint becomes inherent rather than constructed, this leads to a reduced perception of the humorous effect.
6 Summary and conclusion
This paper set itself two objectives. The first one was to describe the family of approximate comparison constructions, whose members follow the syntactic template about as X as Y. It could be shown that the literal and the ironic use are related by subpart links with substantial variations as regards semantic aspects in choosing an appropriate adjective and syntactic as well as semantic variations as regards the Y slot. The choice of the adjective proved to be particularly interesting because some of them are capable of projecting the construct’s interpretation before its completion. Given that, it could be shown that collostructional analyses provide useful tools to uncover such projecting elements. On top of that, the paper showed that some constructions, rather than being monolithic, can be understood dynamically. The second objective of the paper was to draw conclusions from these findings for larger, cognitive accounts of irony. It could be shown that some aspects of the ironic approximate comparison construction raise questions about some of the central assumptions made in the cognitive modelling approach to irony. In particular, this concerns the assumption of an echoic scenario. Studies on the on-line comprehension of the family of approximate comparison constructions could reveal whether the literal interpretation of adjectives which are (statistically) associated with an ironic interpretation is activated, thereby providing evidence in favor of the cognitive modelling approach. The paper showed, though, that the findings on the ironic approximate comparison construction are compatible with (a slightly modified version of) the viewpoint account of irony. It was argued that assuming an inherent viewpoint, which is automatically accessed with the ironic construction, is a necessary amendment to account for the projecting power of the adjective.
Even though only one construction could be considered in this paper, the observations made here raise the possibility of an irony frame. Using elliptical if-clauses as data, Athanasiadou (2020: 82) shows that “irony is evoked due to the alleged incongruity between the source domain of the ‘literal’ world and the target domain of the speakers’ intended world, which may of course involve multiple frames”. These observations bear striking resemblance to the ones that are made on the ironic approximate comparison construction here. However, in the light of the present data, it seems that irony does not only manipulate frames, but may be a frame itself to which the ironic approximate comparison construction is a window. This frame may manipulate other frames in the manner described to achieve an alienating effect on a state-of-affairs by way of viewpoint shift. It might be fruitful for future research to consider such a possibility. Collostructional analyses on other ironic constructions, such as ADJ PRO BE not (Supportive she is not, Giora et al. 2018), could show whether there are general tendencies as regards the properties to be ironicized, which, in turn, provides insights into a possible irony frame.
7 Data availability statement
The datasets generated during and/or analysed during the current study are available in the AboutAsXasY repository, https://seafile.zfn.uni-bremen.de/d/539b328355424c45afe4/.
References
Athanasiadou, Angeliki. 2020. Irony in constructions. In Angeliki Athanasiadou & Herbert L. Colston (eds.), The diversity of irony, 78–90. Berlin/Boston: de Gruyter Mouton.10.1515/9783110652246-004Search in Google Scholar
Auer, Peter. 2009. On-line syntax: Thoughts on the temporality of spoken language. Language Sciences 31(1). 1–13. https://doi.org/10.1016/j.langsci.2007.10.004.Search in Google Scholar
Barsalou, Lawrence W. 1983. Ad hoc categories. Memory & Cognition 11(3). 211–227. https://doi.org/10.3758/bf03196968.Search in Google Scholar
Boyd, Jeremy K. & Adele E. Goldberg. 2011. Learning what not to say: The role of statistical preemption and categorization in a-adjective production. Language. 55–83. https://doi.org/10.1353/lan.2011.0012.Search in Google Scholar
Bredin, Hugh. 1998. Comparisons and similes. Lingua 105(1). 67–78. https://doi.org/10.1016/s0024-3841(97)00030-2.Search in Google Scholar
Cienki, Alan. 2010. Frames, idealized cognitive models, and domains. In Dirk Geeraerts & Cuyckens Hubert (eds.), The Oxford handbook of cognitive linguistics. New York: Oxford University Press.10.1093/oxfordhb/9780199738632.013.0007Search in Google Scholar
Colston, Herbert L. 1999. “Not good” is “bad,” but “not bad” is not “good”: An analysis of three accounts of negation asymmetry. Discourse Processes 28(3). 237–256. https://doi.org/10.1080/01638539909545083.Search in Google Scholar
Coulson, Seana. 2005. Sarcasm and the space structuring model. In Seana Coulson & Barbara Lewandowska-Tomaszczyk (eds.), The literal and nonliteral in language and thought, 129–144. Frankfurt am Main: Peter Lang.Search in Google Scholar
Coulson, Seana. 2015. Frame-shifting and frame semantics: Joke comprehension on the space structuring model. In Geert Brône, Kurt Feyaerts & Tony Veale (eds.), Cognitive linguistics and humor research, 167–190. Berlin/Boston: De Gruyter Mouton.10.1515/9783110346343-009Search in Google Scholar
Coulson, Seana, Thomas P. Urbach & Marta Kutas. 2006. Looking back: Joke comprehension and the space structuring model. Humor 19(3). 229–250. https://doi.org/10.1515/humor.2006.013.Search in Google Scholar
Dancygier, Barbara. 2017. Viewpoint phenomena in constructions and discourse. Glossa: A Journal of General Linguistics 2(1). 1–22. https://doi.org/10.5334/gjgl.253.Search in Google Scholar
Dancygier, Barbara & Eve Sweetser. 2014. Figurative language. New York: Cambridge University Press.Search in Google Scholar
Davies, Mark. 2008–. The Corpus of Contemporary American English (COCA): One billion words, 1990–2019. Retrieved from https://www.english-corpora.org/coca/.Search in Google Scholar
Dynel, Marta. 2013. Irony from a neo-Gricean perspective: On untruthfulness and evaluative implicature. Intercultural Pragmatics 10(3). 403–431. https://doi.org/10.1515/ip-2013-0018.Search in Google Scholar
Fauconnier, Gilles. 1994. Mental spaces: Aspects of meaning construction in natural language. Cambridge: Cambridge University Press.10.1017/CBO9780511624582Search in Google Scholar
Fauconnier, Gilles. 2010. Mental spaces. In Dirk Geeraerts & Cuyckens Hubert (eds.), The Oxford handbook of cognitive linguistics, 352–377. New York: Oxford University Press.Search in Google Scholar
Fauconnier, Gilles & Mark Turner. 2002. The way we think: Conceptual blending and the mind’s hidden complexities. New York: Basic Books.Search in Google Scholar
Fillmore, Charles. 1982. Frame semantics. In Linguistic Society of Korea (ed.), Linguistics in the morning calm, 111–137. Seoul: Hanshin.Search in Google Scholar
Fishelov, David. 1993. Poetic and non-poetic simile: Structure, semantics, rhetoric. Poetics Today 14(1). 1–23.10.2307/1773138Search in Google Scholar
Flach, Susanne. 2017. Collostructions: An R implementation for the family of collostructional methods (R package version 0.1.0). Retrieved from www.bit.ly/sflach.Search in Google Scholar
Garmendia, Joana. 2010. Irony is critical. Pragmatics & Cognition 18(2). 397–421. https://doi.org/10.1075/pc.18.2.07gar.Search in Google Scholar
Garmendia, Joana. 2011. She’s (not) a fine friend: “Saying” and criticism in irony. Intercultural Pragmatics 8(1). 41–65. https://doi.org/10.1515/iprg.2011.002.Search in Google Scholar
Garmendia, Joana. 2014. The clash: Humor and critical attitude in verbal irony. Humor 27(4). 641–659. https://doi.org/10.1515/humor-2014-0094.Search in Google Scholar
Garmendia, Joana. 2018. Irony. Cambridge/New York: Cambridge University Press.Search in Google Scholar
Giora, Rachel & Israela Becker. 2019. S/he is not the most sparkling drink in the pub. global vs. local cue—Which reigns supreme? Metaphor and Symbol 34(3). 141–157. https://doi.org/10.1080/10926488.2019.1649836.Search in Google Scholar
Giora, Rachel, Ari Drucker, Ofer Fein & Itamar Mendelson. 2015. Default sarcastic interpretations: On the priority of nonsalient interpretations. Discourse Processes 52(3). 173–200. https://doi.org/10.1080/0163853x.2014.954951.Search in Google Scholar
Giora, Rachel, Shir Givoni & Ofer Fein. 2015. Defaultness reigns: The case of sarcasm. Metaphor and Symbol 30(4). 290–313. https://doi.org/10.1080/10926488.2015.1074804.Search in Google Scholar
Giora, Rachel, Inbal Jaffe, Israela Becker & Ofer Fein. 2018. Strongly attenuating highly positive concepts: The case of default sarcastic interpretations. Review of Cognitive Linguistics 16(1). 19–47. https://doi.org/10.1075/rcl.00002.gio.Search in Google Scholar
Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago/London: University of Chicago Press.Search in Google Scholar
Goldberg, Adele E. 2006. Constructions at work. The nature of generalizations in language. New York: Oxford University Press.10.1093/acprof:oso/9780199268511.003.0002Search in Google Scholar
Goldberg, Adele E. 2011. Corpus evidence of the viability of statistical preemption. Cognitive Linguistics 22(1). 131–153. https://doi.org/10.1515/cogl.2011.006.Search in Google Scholar
Goldberg, Adele E. 2019. Explain me this: creativity, competition, and the partial productivity of constructions. Princeton: Princeton University Press.10.2307/j.ctvc772nnSearch in Google Scholar
Grice, Herbert P. 1989. Studies in the way of words. Cambridge, MA: Harvard University Press.Search in Google Scholar
Gries, Stefan T. & Anatol Stefanowitsch. 2004. Extending collostructional analysis: A corpus-based perspective on alternations. International Journal of Corpus Linguistics 9(1). 97–129. https://doi.org/10.1075/ijcl.9.1.06gri.Search in Google Scholar
Hoffmann, Thomas & Graeme Trousdale. 2013. The Oxford handbook of construction grammar. Oxford [u.a.]: Oxford University Press.10.1093/oxfordhb/9780195396683.001.0001Search in Google Scholar
Huddleston, Rodney D. & Geoffrey K. Pullum. 2002. In The Cambridge grammar of the English language, (1. publ. ed.). Cambridge [u.a.]: Cambridge University Press.10.1017/9781316423530Search in Google Scholar
Israel, Michael, Jennifer R. Harding & Tobin Vera. 2004. On simile. In Michel Achard & Suzanne Kemmer (eds.), Language, culture, and mind, 123–135. Stanford: CSLI.Search in Google Scholar
Lakoff, George. 1987. Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago press.10.7208/chicago/9780226471013.001.0001Search in Google Scholar
Moon, Rosamund. 2008. Conventionalized as-similes in English: A problem case. International Journal of Corpus Linguistics 13(1). 3–37. https://doi.org/10.1075/ijcl.13.1.03moo.Search in Google Scholar
Ortony, Andrew. 1979. The role of similarity in similes and metaphors. In Andrew Ortony (ed.), Metaphor and thought, 186–201. Cambridge, MA: Cambridge University Press.Search in Google Scholar
Pálinkás, Istvan. 2014. Metaphor, irony and blending. Argumentum 10. 611–630.Search in Google Scholar
Partington, Alan. 2011. Phrasal irony: Its form, function and exploitation. Journal of Pragmatics 43(6). 1786–1800. https://doi.org/10.1016/j.pragma.2010.11.001.Search in Google Scholar
Popa-Wyatt, Mihaela. 2020. Hyperbolic figures. In Angeliki Athanasiadou & Herbert. L. Colston (eds.), The diversity of irony, 91–106. Berlin/Boston: DeGruyter/Mouton.10.1515/9783110652246-005Search in Google Scholar
Romano, Manuela. 2017. Are similes and metaphors interchangeable? A case study in opinion discourse. Review of Cognitive Linguistics 15(1). 1–33. https://doi.org/10.1075/rcl.15.1.01rom.Search in Google Scholar
Ruiz de Mendoza, Francisco J. 2017. Cognitive modeling and irony. In Angeliki Athanasiadou & Herbert L. Colston (eds.), Irony in language use and communication, 179–200. Amsterdam/Philadelphia: Benjamins.10.1075/ftl.1.09demSearch in Google Scholar
Ruiz de Mendoza, Francisco J. & Inés Lozano-Palaci. 2019a. A cognitive-linguistic approach to complexity in irony: Dissecting the ironic echo. Metaphor and Symbol 34(2). 127–138.10.1080/10926488.2019.1611714Search in Google Scholar
Ruiz de Mendoza, Francisco J. & Inés Lozano-Palacio. 2019b. Unraveling irony: From linguistics to literary criticism and back. Cognitive Semantics 5(1). 147–173.10.1163/23526416-00501006Search in Google Scholar
Ruiz de Mendoza, Francisco J., Luzondo Oyón Alba & Sobrino Paula Pérez (eds.). 2017. Constructing families of constructions. Amsterdam/Philadelphia: Benjamins.10.1075/hcp.58Search in Google Scholar
Ruiz de Mendoza, Francisco J & Alicia G. Masegosa. 2014. Cognitive modeling: A linguistic perspective. Amsterdam/Philadelphia: Benjamins.i.10.1075/hcp.45Search in Google Scholar
Stefanowitsch, Anatol. 2013. Collostructional analysis. In Thomas Hoffmann & Graeme Trousdale (eds.), The Oxford handbook of construction grammar, 290–306. New York: Oxford University Press.10.1093/oxfordhb/9780195396683.013.0016Search in Google Scholar
Tobin, Vera & Michael Israel. 2012. Irony as a viewpoint phenomenon. In Barbara Dancygier & Eve Sweetser (eds.), Viewpoint in language: A multimodal perspective, 25–46. New York: Cambridge University Press.10.1017/CBO9781139084727.004Search in Google Scholar
Turner, Mark. 1998. Figure. In Albert N. Katz, Cristina Cacciari, Raymond W. Gibbs & Mark Turner (eds.), Figurative language and thought, 44–87. Oxford: Oxford University Press.Search in Google Scholar
Turner, Mark. 2010. Conceptual Integration. In Dirk Geeraerts & Cuyckens Hubert (eds.), The Oxford handbook of cognitive linguistics, 378–394. New York: Oxford University Press.Search in Google Scholar
Turner, Mark. 2018. The role of creativity in multimodal construction grammar. Zeitschrift für Anglistik und Amerikanistik 66(3). 357–370. https://doi.org/10.1515/zaa-2018-0030.Search in Google Scholar
Veale, Tony. 2013. Humorous similes. Humor 26(1). 3–22. https://doi.org/10.1515/humor-2013-0002.Search in Google Scholar
Veale, Tony & Hao Yanfen. 2010. Detecting ironic intent in creative comparisons. In Paper presented at the 19th European conference on artificial intelligence—ECAI 2010, Lisboa, Portugal.Search in Google Scholar
Wilson, Deirdre & Dan Sperber. 2012. Meaning and relevance. New York: Cambridge University Press.10.1017/CBO9781139028370Search in Google Scholar
© 2021 Claudia Lehmann, published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.
