“I’m gonna get totally and utterly X-ed.” Constructing drunkenness

1 Introduction

The English language is notoriously rich in colourful expressions designating the state of being drunk – i.e. ‘unable to speak or act in the usual way because of having had too much alcohol’ (Cambridge English Dictionary). Common examples of what Shavladze (2023: 1) calls drunkonyms^[1] are the lexical items pissed, hammered, trashed, wasted or trousered.

This is not a recent phenomenon: if we use the advanced search in the Oxford English Dictionary (OED)^[2] to retrieve adjectives containing the words drunk or intoxicated in their definition and sort these by date, we find that merry was already used around 1382 with the meaning ‘Boisterous or cheerful due to alcohol; slightly drunk, tipsy’ (OED).^[3] Another word used very early with the meaning ‘drunk’, but without being a derivative of the word, is blind, with an attestation from 1630:

This metalinguistic quotation from the OED already contains a large number of colourful synonyms, and many more were added to the word-stock of English in the following centuries.

It also appears that the manifold synonyms for drunk raised the interest of word lovers relatively soon: the earliest texts mentioned in the literature assembling this type of expression are John Ray’s collection of proverbs (1678) and Benjamin Franklin’s Drinker’s Dictionary (1737), which contains 228 words and phrases denoting drunkenness, followed by Thomas Norwood’s list of 80 epithets for drunkenness in the Gentleman’s Magazine of 1770 (cf. Crystal 2014: 30; Levine 1981: 1038; Measham 2010: 122). According to Levine (1981: 1038), the 1975 edition of the Dictionary of American Slang listed over 353 synonyms for drunk, making this the concept with the largest number of synonyms according to the editors of the dictionary (cf. also Wentworth and Flexner 1967: xii). In the American Thesaurus of Slang, Levine (1981: 1038) found approximately 900 terms with this meaning – i.e. more than twice as many. The number of synonyms seems to have been on the increase (cf. also Levine 1981: 1050), with the largest figure to be found in a BBC Culture article (Dent 2017) stating that there are “3,000 words English currently holds for the state of being drunk (including ‘ramsquaddled’, ‘obfusticated’, ‘tight as a tick’, and the curious ‘been too free with Sir Richard’)”. While the author provides no supportive evidence for her claim in the form of a word list or a link to a reference work, the figure has been quoted elsewhere and even made it into the game show Funny you should ask.^[4] One may observe that this bears a certain similarity to the inflated numbers reported for snow in “Eskimo” languages – see Pullum (1989) for a discussion.

In view of the prevalence and general interest of the topic (for instance, the audience of BBC One’s Booze programme collected a glossary of 141 euphemisms for being drunk; cf. BBC Booze Programme 2002), it is highly surprising that drunkonyms are still under-researched from a linguistic perspective. Besides the aforementioned (quasi‑)lexicographical treatments of the topic, only few studies are concerned with synonyms for drunk. Furthermore, these limit themselves to short lists of synonyms produced by groups of researchers or staff working in the field of alcohol treatment (Cameron et al. 2000; Levitt et al. 2009), with the aim of providing more accurate ways of talking about degrees of and attitudes towards drunkenness in English and other languages. Crystal (2014: 30) provides a semantic analysis of the word field, suggesting that the colourful words for drunk seem to refer to aspects such as the appearance of the drunk person (e.g. pie-eyed), their erratic or lacking movements (e.g. looped, stiff) or their mental state (e.g. fuddled), but it only comprises a comparatively small number of items.

This contribution thus fills a gap in linguistic research by providing a detailed investigation of the linguistic means that present-day English uses to talk about drunkenness. While the state of drunkenness and alcohol consumption in general are linked to many risks and problems (cf. e.g. World Health Organisation 2023), they seem to have played and do still play an important role in various societies or cultures, with a certain degree of acceptance and more positive attitudes towards heavy drinking in the cultural and historical context of certain festivities (cf. Measham 2010). The mere number of synonyms for drunk in the English language testifies to the cultural significance of this concept in English-speaking societies (cf. Measham 2010: 122 and Barr 1995; again, this is a parallel to the debates about words for snow). We therefore agree with Levine (1981: 1050) that the “ever-increasing number of expressions and terms about drunkenness […] deserve to be considered seriously” and will investigate the linguistic means employed in the context of the ideally “nonpathological drunkenness” of “celebrations, of beginnings and endings” (Levine 1981: 1038). We definitely do not wish to promote the state of drunkenness in view of its manifold negative consequences but approach the issue from an exclusively linguistic perspective.

As a starting point for our research, we compiled a list of synonyms for the adjective drunk, which combines material extracted from fairly recent user-generated word lists (BBC and Wiktionary) with material from a thesaurus by a dictionary publisher (Collins) and items extracted from the largest dictionary of the English language (OED), using the following resources:

1. the results of the aforementioned OED searches for adjectives whose definition contains drunk or intoxicated (30th October, 2023). These were manually checked for semantic compatibility with alcohol-related intoxication.^[5]

2. the list collected by the audience of BBC One’s Booze programme (2002) referred to above

3. a list of British English synonyms for drunk from Collins Dictionary’s thesaurus (4th May, 2023)

4. a list of synonyms for drunk from Wiktionary’s thesaurus (4th May, 2023).

We excluded synonyms including drunk as a base (e.g. drunky or martin-drunk) as well as phrases like Mozart and Liszt or worse for drunk. We then used Excel to automatically delete repeated items. The resulting list of 546 synonyms for drunk can be found in Appendix 1.^[6]

A very light-hearted and entertaining argument why English has so many words referring to drunkenness is suggested by British stand-up comedian Michael McIntyre. In one of his famous comedy routines (2)^[7], he claims that “posh people” can use any word in the English language to mean ‘drunk’:

In a variation of his standup routine (McIntyre 2011), Michael McIntyre uses slightly different examples and points out that “it still sounds acceptable”. He adds trolleyed to the list of colourful synonyms for drunk and begins with the example “I got utterly gazeboed”, followed by “Angus, you’re driving, I’m carparked” and “Slow down, chap, I’m pyjamaed”.

Only few of the convincing examples provided by McIntyre are attested in the Oxford English Dictionary (trousered, arseholed, rat-arsed, trolleyed). This would suggest that the humour of the passage derives to a large extent from McIntyre’s own, more or less spontaneous innovative word formations, whose bases additionally come from very disparate domains; e.g. wellied, gazeboed, carparked and pyjamaed. Interestingly, however, it is possible to find Urban Dictionary entries for all of these items, which date from 2006 (wellied^[8]), 2008 (Gazeboed^[9]) and 2009 (carparked^[10] and Pyjamad^[11]), respectively. This means that these terms were already in use either shortly before or around the time of McIntyre’s tour, which was recorded in October 2009 to be published as a DVD^[12]. Still, the strong reactions of McIntyre’s audience, who respond with much laughter and clapping, seem to imply that these words are unfamiliar to a large proportion of the listeners. Their understanding of the words’ meaning would thus rely on the previous context, in which the colourful synonyms of drunk were presented.

If McIntyre’s hypothesis were right that any English word can be used to mean ‘drunk’,^[13] and if a language user wanted to apply this principle, it would be necessary to ensure the listeners’ understanding of the innovative use of the word either through contextual clues or other linguistic means. Based on the examples that McIntyre provides, we can provisionally narrow down the claim on formal grounds to linguistic items that constitute a single chain of letters (with the possible insertion of a hyphen) and which end in an -ed suffix. This restriction can be justified by the fact that 312 of our 546 drunkonyms in Appendix 1 contain an -ed suffix, either at the end of the word (e.g. plastered) or inside in those cases where it ends with a particle (e.g. tanked up). This affix corresponds to the suffix -ed² described in the OED, which can be added to any noun in order to form an adjective (without the need to have an intervening verb for past participle formation, as with the OED suffix -ed¹). The resulting adjective usually has the meaning ‘possessing, provided with or characterized by NOUN’, but the OED also mentions adjectives like dogged ‘obstinate’ or crabbed ‘cross-tempered’, where the meaning of the suffix is vaguer. Even in these cases, though, it would seem that there is usually at least some metaphorical or metonymic motivation for the use of the base (e.g. by relating the unusual sideways gait of crabs to contrarity). This is also the case in the linguistic accounts attempting a semantic subcategorization of drunkonyms.^[14] By contrast, it is highly difficult if not impossible to establish any semantic relation between the nouns forming the base of McIntyre’s humorous terms for drunk and the meaning ‘drunk’, and to determine in what way intoxication is like a gazebo. One may even venture to claim that the more distant and less obvious the relation between the base and the meaning ‘drunk’, the better the word fulfils its humorous goal; possibly because this discrepancy reflects the disorderly state of drunkenness.

As a consequence, linguistic means might need to play a more important role to secure understanding in this context than in other instances of word formation. What is striking in McIntyre’s examples is that he very frequently uses intensifying premodifying adverbs before the humorous adjectives referring to drunkenness: utterly gazeboed (twice), totally and utterly carparked and fucking pyjamaed. This is in line with Crystal’s (2014: 30–31) observation that drinkers use certain terms to “exaggerate their state during or after a drinking session, usually by suggesting the thoroughness of their achievement” and by expressing that some maximum has been reached, which results in a state of “being ‘completely filled’” (Crystal 2014: 31).

As for the types of verb used in McIntyre’s examples representing fictional dialogues, we overwhelmingly find instances of BE, followed by some uses of GET. All example sentences use the first person and follow a predicative pattern. The tense is either past or present, and McIntyre also uses one going to future. Interestingly, all references to the past^[15] comprise a premodifying adverb (see above), whereas all references to a present state of drunkenness do without it:

One may argue that the extra-linguistic context of the sentences referring to the present in the first person singular will already provide clues about the speaker’s state of drunkenness to the listener. Additionally, both (3) and (4) begin with an introductory part providing further hints (highlighted in the examples). Where this is not the case, by contrast, adverbial premodifiers seem to constitute an important aspect of talking about drunkenness in colourful words. All this seems to suggest that the pattern “I BE/GET ADV X-ed”, or more specifically “I BE/GET utterly X-ed” is most prototypical when talking about drunkenness in a humorous way.

In order to determine whether this observation can be generalised or whether this is an idiosyncratic use in a comedy routine, we are going explore both established and less established ways of expressing drunkenness in corpora and in lexical resources such as dictionaries and thesauri in the remainder of this paper. We will introduce and discuss the results of an empirical study and propose a constructionist treatment of drunkenness, which has been sorely missing from Construction Grammar theory so far.^[16]

2 Theoretical questions

If McIntyre is right that “any word in the English language” can be used in a predicative sentence structure to mean ‘drunk’, this would suggest that the semantic component of drunkenness is not stored as a property of the respective word itself, very similarly to the situation described by Goldberg (1995, 2006) for the ditransitive construction. Goldberg shows that even when novel verbs are used within the construction, the notion of transfer is still activated and thus needs to be contributed by the Argument Structure Construction (ASC). Example (5) illustrates how the clausal configuration carries the meaning of ‘transfer’ when crutch (which usually occurs as a noun) is used in the verb slot of an ASC.

Accordingly, we would expect to find a larger construction in contexts where language users talk about drunkenness, which allows us to interpret whichever word occurs within it as expressing the state of being drunk.

Such a constructionist treatment is supported by various observations:

1. The number of word types that can be used to mean ‘drunk’ is extremely large: thus Larson quotes from Benjamin Franklin, who, in his 1737 article listing numerous synonyms, does “not doubt but that there are many more in use” (cf. Larson 1937: 92), and Wentworth and Flexner (1967: 716) find that “slang synonyms for drunk continue to appear with amazing frequency”.

2. At the same time, many of the terms “are represented by just a single citation” (Crystal 2014: 30), and they fall out of use relatively quickly. Thus Levitt et al. (2009: 448) note that “some of the terms Levine offered as current in 1981 (e.g., ‘bagged’ and ‘twisted’) are anachronistic today”.

If the situation is similar to that of the ditransitive construction and many other constructions, then we would also expect to find a skewed distribution of the items occurring in the construction. For instance, the verb give amounts for roughly half of the occurrences of the ditransitive construction (Goldberg 1995, 2006; Stefanowitsch and Gries 2003), and its semantics matches the meaning of the construction as a whole. According to Goldberg (2006: 75–85), this “skewed input” helps in the acquisition of the construction. Ellis and Ferreira-Junior (2009) show similar “Zipfian” distributions for a range of their Verb-Argument-Constructions, e.g. for go in Verb-Location or for put in Verb-Object-Location, where many established and certainly all novel and creative uses account for only a small fraction of the overall occurrences of the construction. For a potential drunkenness construction, we would therefore expect either drunk or its most frequent humorous synonym to act as the most frequent and semantically most compatible element, dominating the competition in its slot.

Since the structures with be and get mentioned in the previous section are semantically different, this raises the question whether we should expect to find a smaller construction that is used in both common patterns, or whether<thinsp>there are at least two drunkenness constructions, one with be and one with<thinsp>get.

The theoretical questions we would like to address in this contribution can thus be summed up as follows:

1. Can really any word be used to express the notion of drunkenness in English?

   1. If so, what are the conditions under which that is possible?

   2. If not, what restrictions are there?

2. Is it helpful to model such euphemistic uses constructionally…

   1. … in the form of a schematic drunkenness construction and/or

   2. … in the form of multiple, less abstract drunkenness constructions?

To answer these questions, we carried out an empirical corpus study, which will be described in the following.

3 Material and method of the corpus study

If we assume that any word can be used to mean ‘drunk’, this poses a methodological problem for corpus research, because there are no specific word(s) to query. Most of McIntyre’s examples occur in structures with the lemmas be and get, but these are two of the most frequent verbs of the English language, with frequencies of over 4 million (be) and over 200,000 (get) respectively in the 100-million-word 1994 British National Corpus (BNC). Our approach was therefore to first search for these two lemmas followed by optional elements (to allow for adjuncts but also for premodifiers) and then by the prototypical known word(s), e.g. drunk. This enabled us to identify relevant patterns that we could then use in a second step to query the corpus without the known word(s).

More specifically, we used BNCweb (Hoffmann/Evert 2006) to search the BNC for the patterns

drunk_{A}<<5<<{be}

drunk_{A}<<5<<{get}

i.e. for the form drunk tagged as an adjective following the respective lemmas be and get with a maximum of 5 intervening words. Since we expected the pattern to occur particularly frequently in spoken conversation, we also searched the 11.4-million-word Spoken BNC 2014 (Love et al. 2017) using the same queries as above in CQPweb (Hardie 2012). Results were sorted by the element to the left of drunk and then downloaded for further analysis in a spreadsheet.^[17]

Our search retrieved the target structures, as in (6) and (7), but also instances of erroneous part-of-speech tagging, as in (8), where drunk is used as a noun rather than an adjective, or examples like (9), in which the verb be does not relate to the adjective drunk. Similarly, example (10) is a false hit for be (but was also correctly retrieved for get).

Within the BNC, we expected to find the target structures more frequently in spoken conversation (i.e. the spoken demographically-sampled portion) than in more formal spoken texts such as radio interviews and lectures (i.e. the spoken context-governed portion). Accordingly, we queried both of these spoken subcorpora and the written subcorpus separately.

In the next step, all premodifiers of drunk were extracted manually to find patterns that would help us identify novel uses. We ignored duplicates and only considered items that undoubtedly premodified drunk, thereby excluding adverbs such as already, always or actually, e.g. in (11):

To allow for a more automated analysis of the data, more restricted queries were used for a collostructional analysis, specifically for simple collexeme analysis in the sense of Stefanowitsch and Gries (2003). We ran two queries on both corpora (BNC 1994 and Spoken BNC 2014) and joined the concordances of both corpora: the first query was for be or get as lemma, followed by an adverb, followed by an adjective. With this query, we captured established adjectives like drunk. However, rare or novel forms like shitfaced or plastered are often erroneously tagged as verbs of various types (e.g. past tense forms or past participles) and would thus have been impossible to retrieve in this way. We therefore used a second query for be or get as lemma, followed by an adverb, followed by a form ending in the orthographic string of characters <ed> to capture more innovative items. The results of both queries were reduced to 2-grams consisting of the premodifying adverb and the adjective (346,852 hits) or -ed form (145,133 hits). For the calculations, we used these raw frequencies as the number of all constructions, following Proisl (2019). All other fields in the contingency table were filled as described by Stefanowitsch and Gries (2003) and as is common in collocation research (see Evert 2005 for details). Table 1 illustrates this with the example of the combination really drunk, which occurs 39 times in our sample. The premodifier really on its own occurs 12,731 times in total, and thus 12,692 times without drunk. The adjective drunk occurs 358 times, i.e. 319 times without really. We then applied Fisher’s exact test to this table to determine whether really and drunk occur more frequently together than expected (which they do – see Section 4).

For the simple collexeme analysis of the set of -ed drunkonyms, we followed the same protocol as for individual items but pooled all drunkonyms together, so that the contingency table contains frequencies for all drunkonyms with a given premodifier instead of just one specific one, as in Table 1.

4 Results and discussion

5 Conclusion

In this paper, we have provided further evidence that there is – and has been for a long time – a wealth of terms for states of drunkenness in the English language. We have also shown that there are (at least until recently) productive patterns of creating new words meaning ‘drunk’, or rather ‘pissed’. A meaningful argument structure construction without any lexical content might be available to speakers, but we conjecture that additional contextual cues as to the topic of drunkenness will be needed to successfully use this construction, the most prototypical form of which is

be/get + intensifying premodifier + -ed-form.

Coming back to the theoretical questions asked at the end of Section 2, we can say that the wide range of words observed in the already existing lists of drunkonyms seems to support the view that there is a large amount of words that one could potentially use to creatively express drunkenness in English. The wording “any word” put forward by McIntyre appears slightly too general, though, as it is difficult to imagine words such as is, the or of to mean ‘drunk’. Also, for nouns such as carpark or gazebo, it is strictly speaking not the word itself but an -ed-form of it, i.e. carparked or gazeboed, that expresses the meaning of drunkenness in the relevant contexts.

The question of whether it makes sense to model such uses constructionally in the form of a schematic drunkenness construction and/or in the form of multiple, less abstract drunkenness constructions can be answered in the affirmative for both. There are at least two partially productive drunkenness constructions in English, the one given in the previous paragraph, and the rhyming slang approach briefly mentioned at the end of Section 4. Possibly, there is a third one with nicely and blind, but due to the morphological difference between the two, this would have to be a rather high-level construction, and the generalization it provides is rather limited.

There is ample potential for future studies into the question of expressing drunkenness in English, as our two productive constructions only apply to less than half of all drunkonyms collected in Appendix 1. Thus forms such as blotto, slug-nutty or stocious need accounting for, and it appears that in the expression of drunkenness in English, the storage of seemingly unmotivated forms plays a major role. Construction Grammar, with its strong focus on storage, is highly suited to model all the linguistic aspects of English drunkenness.