1. Introduction
Strong language originating from taboo expressions is interesting for a variety of reasons. It flavors our speech, it shows great variation among social groups and especially social settings, and it changes all the time. In the past, lexicographers, sociolinguists, folklorists, anthropologists and historical linguists have occasionally shown an interest in the topic (cf. e.g. Pott Reference Pott1833, de Jager Reference de Jager and Arie de1858, de Baere Reference de Baere1940, Cameron Reference Cameron1969, Andersson & Trudgill Reference Andersson and Trudgill1990, Allan & Burridge Reference Allan and Burridge1991, Dundes Reference Dundes2002 and much of the contents of the journal Maledicta), but only relatively recently have theoretically inclined linguists begun to pay attention to this matter (e.g. Brame Reference Brame1978; Pesetsky Reference Pesetsky, Eric and Alice1987; Postma Reference Postma1995, Reference Postma and Hoeksema2001; Horn Reference Horn and Hoeksema2001; Hoeksema Reference Hoeksema and Laurel2001, Reference Hoeksema2002; den Dikken & Giannakidou Reference den Dikken and Giannakidou2002; Merchant Reference Merchant, Jan-Wouter and Werner2002, Reference Merchant, Martin and Henk van2006; Huang & Ochi Reference Huang, Ochi, Kier and Wolf2004; Sprouse Reference Sprouse, Aviad, Tatjana and Joshua2005). We think it is only fitting that they should do so, for the data are not just colorful, but also instructive on a number of points.
The main goal of this paper is to analyze and compare the two English constructions exemplified in (1) and (2), respectively:
- (1)
(a) Let's get the hell out of here.
(b) Get the hell off my property.
(c) Shut the hell up.
(d) Leave her the hell alone.
(e) Why don't you go the hell away.
(2) I {beat/kicked/annoyed/punched/surprised/irritated} the hell out of him.
We show that these constructions, despite their surface string similarity, differ syntactically and semantically, and that their analysis is best characterized in terms of constructions as building blocks. An interesting by-product of this study is the light shed on historical change in general: taboo terms are shown to extend their distribution to varying syntactic structures in a parallel way. This suggests that a pragmatically unified but semantically disparate class of expressions can be treated as a unit by the grammar.
2. A remark on the corpus
Certainly a study of the syntax of the two constructions at issue could be carried out strictly by tapping into native speakers' intuitions, which we have done to a large extent. However, in order to delve into the history of these constructions, we needed access to a written corpus.
We therefore base our discussion in part on a corpus investigation, largely using material from scanned books and newspapers available through the Google website. In addition, we used online newspaper material from LexisNexis, as well as e-texts from the Gutenberg project, and material from various websites accessed through Google.
Reasons for using this material, rather than existing diachronic corpora, include the need for size (many books do not contain a single instance of the constructions we are interested in, so we needed a big corpus), and the possibility, offered by Google Books and the archive part of Google News, to search for books and newspapers from a particular period (say, material published between 1850 and 1900). We have used this material with caution, fully aware of the many scanning errors it contains, deciding not to use material when we doubted its true publication date. From these various sources, we collected a total of 1000 occurrences, judging that to be sufficient to paint a reasonably accurate picture of the distributional and historical trends that we are interested in.
3. Taboo terms and their linguistic exploitation
Taboo terms come from a number of sources and are used in a wide variety of domains. We will briefly list the main sources for English, and then discuss the various ways in which the taboo terms are used. The purpose of this section is to show that the class of taboo terms is semantically diverse, but pragmatically coherent.
There are three major sources of taboo terms in English. Many terms, especially older ones, stem from religious and folk beliefs, such as those having to do with the devil, hell, God, Jesus Christ, saints, heaven, salvation and damnation. Sexual acts and reproductive organs also provide strong linguistic taboos, even today, in spite of the sexual revolution of the 1960s and 1970s. Finally, bodily excretions constitute powerful taboo topics, and provide us with an additional set of taboo terms to choose from (and see Horn Reference Horn2004 for an exposition of the bodily fluid imagery behind the expression spitting image).
Other possible taboo topics, such as money, though compared to feces by Freud (Reference Freud1913), do not seem to have been used in strong language of the type considered here, nor do the group of life, death, and diseases except in name-calling (You scrofulous idiot!, and see Potts & Roeper Reference Potts, Roeper, Ljiljana, Kate, Eugenia and Ellen2006), although other languages make wider use of this group (Van Sterkenburg Reference Van Sterkenburg2001, Kehayov Reference Kehayov2006). Race is also a taboo topic in polite conversation, but it, like money, is not a source for strong language in English, apart from epithets used to put racial and ethnic groups down. Extreme political or social positions can be taboo among certain people as well, and can give rise to epithets, such as the general bigot or the more specific feminist – it all depends on the speaker's point of view. We also can call people animals in some of the same kinds of situations in which we employ taboo terms, but few of these terms have the force of a taboo (an exception being bitch).
Taboo terms are used in many ways. We will discuss only some, because a more complete discussion would take us too far far afield. The best-known use of taboo terms is probably in exclamations, i.e., cursing and swearing, as in (3a) and the quasi-curses that stand for the real McCoy in (3b):
- (3)
(a) Damn!/Fuck!/Shit!
(b) Darn!/Fudge!/Shoot!
Name-calling is another frequent use:
(4) You {devil/cumdumpster/turd}!
Closely related to swearing and name-calling are maledictions, often taking the form of imperatives:
(5) Go to hell!/Suck a hairy moose cock!/Eat shit and die!
Taboo terms can be used as predicates with non-literal meanings:
(6) That sucks!/She shits on us all the time!
Often particle verbs of taboo origin form rude expressions for telling people to go away:
(7) {Fuck/Piss} off!
They can be used as pejorative modifiers:
(8) the professor from hell/a dick thing to do/a shitty proposal
Some taboo terms develop into degree adverbs:
- (9)
(a) He's hell-bent on doing it./You're so damned clever.
(b) He's butt ugly./That's fucking awesome.
(c) She's spitting mad./He's rabidly insane.
Notice that fucking in (9b) functions as an intensifier, in contrast to its use elsewhere as a pejorative modifier (What a fucking dickwad). We note in passing that cases like hell bent or butt ugly are elative compounds, but they can nonetheless be viewed as involving intensification by modification in exactly the same way as more clearly phrasal cases of adverb+adjective. Still others are used in similar ways in resultative and comparison constructions to indicate a high degree:
(10) He was funny as hell./We were scared shitless.
Taboo terms can express emphatic rejection or denial, as well as emphatic assertion. The denial or negative contradiction construction contains no overt standard negation element, but uses a preposed taboo term instead:
(11) {The/Like} {hell/fuck} I will.
Emphatic assertions with taboo terms, however, are not restricted to a single construction type:
- (12)
(a) You bet your {(sweet) ass/booty} I am.
(b) Damn {right/straight}.
(c) Sure shit.
(d) Does the bear shit in the woods?
(e) I'm not shitting you.
It is also possible to put a taboo term in front of yes, no and yeah:
(13) Hell yes!/Fuck no!/Shit yeah!
Both denials and assertions of these types are appropriate only as a response to another statement, and may not be syntactically embedded:
- (14)
(a) A: I hear you're going to marry a Norwegian woman.
B: (*I believe that) The hell I will.
(b) A: I hear you're going to marry an Italian woman.
B: (*I believe that) You bet your sweet ass I am.
English wh-questions (including rhetorical ones) may be made more emphatic by means of various taboo terms (as well as some free relative clauses: I will do whatever the hell you do, although not all free relatives permit the intrusion of expletive elements: *I will do what the hell you do; Fillmore Reference Fillmore, Mary, Mary, Vassiliki and Deborah1985: 81):
- (15)
(a) {Who/Where/What} the {hell/fuck} are you?
(b) Why the hell should I care?/What the shit is going on?
Important to note for our later discussion is that some taboo terms are used as negative polarity items (cf. Horn Reference Horn and Hoeksema2001, Postma Reference Postma and Hoeksema2001, Hoeksema Reference Hoeksema2002, Postal Reference Postal2005). Subsets can be distinguished, such as minimizers with a pejorative adjectival modifier of taboo origin:
(16) I can't see a {damned/fucking/bloody} thing.
as well as mass nouns (sometimes with an added particle all):
(17) They didn't {say dick/know jack shit} about it.
Terms in the latter, but not the former, set may also appear without any form of overt negation, while still carrying a negative sense (Horn Reference Horn and Hoeksema2001, Postal Reference Postal2005), which means they have undergone a turn of the Jespersen cycle (Jespersen Reference Jespersen1917):
(18) They did {dick/shit} about it, that's what they did.
A third set of negative polarity expressions are of the form give an X, where X is typically, though not exclusively, some taboo term:
(19) give a {damn/hoot in hell/(flying) fuck/rat's ass/crap/shit}
Most of these can appear without the negative while still carrying the negative sense, parallel to the examples in (18).
The above examples of various constructions employing taboo terms are meant not only to illustrate the use of taboo terms, but also to make a more serious linguistic point. The constructions vary widely, and have very little in common, apart from the fact that they all appear to have an emotionally charged character. In these constructions, the taboo terms behave like syntactic silly putty that can be bent and shaped every which way. Normally, the enlistment of lexical items by a construction is based on the category and features of the items in question, and these in turn are in part determined by lexical semantics, and in part arbitrary. In taboo constructions, however, lexical meaning appears to play no role. If anything, these constructions show the victory of connotation over denotation. The fact that taboo terms have a certain rude quality about them is more relevant than their meaning. This is quite obvious when we consider the many constructions where fuck may be used instead of hell. Semantically, the two words are rather different, apart from their taboo status. We address the theoretical import of such spreading in section 6.
4. Two constructions
The two English taboo-term constructions analyzed in this paper have identical surface strings: verb+taboo term+PP (although the final part, the coda, can vary). The first we call get-the-hell-out, or the G-construction, as seen in (1) above and in (20):
(20) Let's get the hell out of this cow town./Back the hell off!
The second we call beat-the-hell-up, or the B-construction, as seen in (2) above and in (21):
(21) They're beating the hell out of Jones./She scares the hell out of me.
Evidence that they are syntactically distinct follows.
4.1. Object status of the taboo term
The taboo terms in the B-construction are direct objects, whereas their counterparts in the G-construction are not. For one thing, taboo terms cannot be omitted from the B-construction, whereas they can from the G-construction:
- (22)
(a) They scared *(the shit) out of me.
(b) They got (the hell) out of the car.
Additional evidence comes from the possibility of passivization. While ordinary passives are at best marginal in the B-construction (cf. (23d)), get and have passives (cf. (23b–c)) are fine:
- (23)
(a) They kicked the shit out of Jones.
(b) Jones got the shit kicked out of him (by them).
(c) Jones had the shit kicked out of him (by them).
(d) ?The shit was kicked out of Jones (by them).
Compare with the poor results of trying to do the same with the G-construction:
- (24)
(a) They stayed the hell away from John.
(b) *John got the hell stayed away from him.
(c) *John had the hell stayed away from him.
(d) *The hell was stayed away from John.
(Cf. They stayed the execution again vs. The execution was stayed again.)
Correlating in a loose way is the transitivity status of the verbs in the B- and G-constructions. In our collection of 1000 natural occurrences of constructions B and G, we found the distribution of transitive and intransitive verbs as presented in table 1. (The numbers in the table don't add up to 1000 because of 15 cases not involving verbs, which we omit here. We discuss these cases in section 4.7 below.)
Table 1 Transitivity and construction type
The correlation is striking, and definitely significant, but not absolute.Footnote 2 This is due, in part, to the fact that G-construction sentences like (25a, c) may have transitive counterparts such as (25b, d):
- (25)
(a) Let's get the hell out of the car.
(b) Let's get him the hell out of the car.
(c) Let's run the hell out of town.
(d) Let's run 'em the hell out of town.
In these transitive cases, it is even clearer that the taboo term the hell is not a direct object, since the object position is occupied here by the pronouns him and 'em respectively. Again, passivization makes the point:
(26) Those cattle rustlers {were/got} run the hell out of town.
(Since passivization of main-verb get yields poor results, independent of the taboo constructions, we do not provide a passive counterpart to (25b).)
Another factor involved in making the correlation in table 1 less than perfect is the existence of intransitive verbs partaking in the B-construction. The four examples in our collection are:Footnote 3
- (27)
(a) She can walk the shit out of a lengthy platform …
(b) You make me laugh the shit out of myself …
(c) I used LimeWire Pro to download porn and masturbate the hell out of my dick …
(d) I sat in the waiting area, shaking my foot, fixing my shoelaces, basically fidgeting the hell out of myself.
Again, these putative exceptions corroborate our proposal when looked at more closely. Intransitive verbs in general can occur with cognate objects, as in (28).
(28) She slept the sleep of the weary.
Here slept and sleep are morphologically related. But such objects need not be actually morphologically related to the verb; so long as the object represents the referential extension of the action of the verb, it is acceptable:
(29) She cried crocodile tears./It rained cats and dogs.
In these instances, the direct object serves to add information to the manner of the action of the verb. That is, we don't learn that the action was done to a particular thematic argument; rather, we learn that the action proceeded in a particular way. Generally the manner information we get concerns the degree of the activity. In sleeping the sleep of the weary, we know that the sleep was long and heavy. In crying crocodile tears, we know that this was a noisy, showy action.
Functionally similar fake DOs can be licensed by a resultative:
- (30)
(a) I am going to walk these shoes *(to pieces).
(b) We will laugh ourselves *(silly).
Resultatives do not always have a literal interpretation; instead, some may be used as degree expressions. For instance, starving to death may literally be dying caused by starvation; but being bored to death does not usually mean one is dying as a result of boredom, but, rather, denotes a high degree of boredom. It is with this type of interpretation in mind that we see the examples in (27) as having fake objects, akin to those in both cognate-object constructions and resultatives. Walking the shit out of a lengthy platform (as a model might) means walking that platform very well or very often, and beating the hell out of someone is not to cause the hell to come out of someone by means of rather forceful exorcism, but to beat someone a lot, to a high degree of intensity.
If we think of the relation between beat Harry and beat the shit out of Harry as a verb alternation (à la Levin Reference Levin1993), it is clear that it has the following properties: the regular DO of beat is turned into the object of a preposition (out of) and the place of the DO is now taken up by a definite NP involving some taboo N: the hell, the shit, etc. The object of the preposition is still interpreted as the theme or patient of whatever action is described by the verb, while the entire construction indicates a high degree of intensity for the action in question.
4.2. Polarity issues
Den Dikken & Giannakidou (Reference den Dikken and Giannakidou2002) argue that what the hell and other reinforced wh-items are polarity items of a particular sort. When embedded under a predicate such as know, which is not specifically a question predicate, they are best when know in turn appears in a nonveridical environment. A nonveridical environment for some proposition p is one that does not permit one to infer the truth of p (Zwarts Reference Zwarts1995, Giannakidou Reference Giannakidou1998). Examples of nonveridical environments are complements of intensional verbs, the scope of negation, conditionals, questions, modal contexts, imperatives, etc. Compare:
- (31)
(a) ?I know where the hell he lives.Footnote 4
(b) I don't know where the hell he lives.
(c) I want to know where the hell he lives.
(d) Pray tell me where the hell he lives.
(e) Do you know where the hell he lives?
(f) Only his parents know where the hell he lives.
The G-construction, on the other hand, is a positive-polarity construction:Footnote 5
- (32)
(a) *Don't get the hell off my property!
(b) *Jones did not get the hell out of Dodge.
(c) *Let's not get the hell out of here.
Note that the examples in (32) are more acceptable with contrastive or echoic negation, something which is generally true of positive polarity items (see Seuren Reference Seuren1985, Horn Reference Horn1989, among others). This may be hard to show in the case of (32a), which is an imperative, but (32b) could be used in a context such as:
(33) Jones was told to get the hell out of Dodge, but Jones never does as he is told. So he did not get the hell out of Dodge.
The same is true for constituent negation. In (34) not negates the adjunct for nothing, but leaves the main assertion intact (i.e. Jones got out of Dodge).
(34) Jones did not get the hell out of Dodge for nothing.
Also, it is well-known that when negation is embedded in questions, conditionals, or other negative contexts, yielding a double negation context, the occurrence of positive polarity items is often grammatical (Baker Reference Baker1970, Horn Reference Horn1989, Szabolcsi Reference Szabolcsi2004); compare:
- (35)
(a) I'll beat you if you don't get the hell off my property!
(b) Why don't you get the hell off my property!
(c) You can't say he did not get the hell out of here.
It may appear odd that the hell sometimes produces negative polarity items and sometimes positive polarity items, but it is actually not uncommon for lexically similar expressions to diverge in this way (Hoeksema Reference Hoeksema1996).
The B-construction, unlike the G-construction, is not a positive polarity item, and may appear with ordinary negation, without any need for special echoic intonation or double negation contexts:
- (36)
(a) We did not beat the hell out of him.
(b) AIDS does not scare the shit out of people anymore.
(c) I don't straighten the hell out of my hair anymore; curly is in.
(d) Nothing could depress the hell out me; I'm an optimistic sort
4.3. Choice of verbs
The two constructions are clearly distinguished by the verbs they enlist. As is apparent from table 2, the most common verbs in the B-construction involve some kind of physical abuse verb, such as beat, kick, knock, shoot, smash, sock, etc. In addition, a large group of psychological verbs also can be used in this construction, some of which, such as scare and frighten, appear quite frequently. Mostly, verbs in the B-construction have a negative connotation, but more positive ones also show up, such as impress.
Table 2 Verbs in the B-construction
In the special case of beat, we note that the expletive may equally combine with this verb when it is used literally, or when it is used as part of the idiom that beats (the hell out of) me (‘I don't have a clue’). On the other hand, the idiom beat it (‘go away’, as in the eponymous Michael Jackson hit song) does not seem to occur in the B-construction. We assume that the verb alternation involved requires the object of the complex preposition out of to be able to bear a thematic role. This will prevent verbs with dummy objects from undergoing the alternation. Compare also I'm gonna cut the hell out of this dress with Good grades alone won't cut (*the hell out of) it anymore in the competitive world of college admissions.
The large number of verbs occurring just once in our data points to a considerable productivity of the B-construction.Footnote 6 One need only google the fixed string the hell out of to find many more verbs in this construction on the Internet. It is not our aim to provide complete lexicographical coverage of the construction, but rather to give a representative sample of the kinds of verbs involved, as well as which ones are most commonly used.
The set of verbs used in the G-construction (table 3) shows almost no overlap with the set of verbs listed in table 2.
Table 3 Verbs in the G-construction
It may come as a surprise that the G-construction, though slightly more common than the B-construction, is actually less productive, in that it shows less variety in the choice of verbs. In interpreting table 3, the reader should bear in mind that some of the verbs are part of a verb-particle combination, where we lump particles with intransitive prepositions (Emonds Reference Emonds1976: 174). So shut, for instance, is found frequently because of the particle verb shut up, as in Shut the hell up! Similarly, leave is a frequent verb in the list because of leave alone, cf. Leave me the hell alone! Since the most common meaning component of the predicates in the G-construction is movement, it may seem odd to list shut up or leave alone under the rubric of the G-construction. Clearly, this decision needs to be motivated, and we will do so in the next section. In the mean time, we note that none of the verbs listed in table 2, when occurring in the B-construction, are part of particle verbs.Footnote 7
4.4. Choice of taboo terms
In table 4 we show the frequencies of the various taboo terms that we have found in our data set, dividing them over the two constructions G and B. Fifteen special cases that did not clearly belong to either G or B were left out of the table. See section 4.7 for discussion.
Table 4 Taboo terms in the G- and B-constructions
We see in table 4 that the B-construction not only shows greater productivity when it comes to verbs, but also with regard to taboo terms. In fact, the majority of taboo terms found in the B-construction do not appear in the G-construction, nor are they legitimate there, according to our intuitions. While some of the items listed (e.g. the living pudding) are clearly one-off cases, usually found on the Internet and showing creative rather than conventionalized use of language, most are robustly attested.
Compare, for instance, (37) and (38):
- (37)
(a) Let's get the {hell/*shit/*bejesus} out of Dodge.
(b) Shut the {fuck/hell/*shit/*bejesus} up.
(c) Leave her the {fuck/hell/heck/*shit/*bejesus} alone.
- (38)
(a) They beat the {hell/fuck/shit/bejesus} out of him.
(b) The police kicked the {hell/fuck/shit/bejesus} out of them.
(c) That scares the {hell/fuck/ shit/bejesus} out of me.
The asterisks in (37)–(38) reflect the native-speaker judgments of student and faculty informants at Swarthmore College in the early months of 2006. Not all speakers have such a clear distinction, as one of our reviewers pointed out to us. However, it would appear that the Swarthmore academic community we tested is, in fact, revealing a tendency that others share. On 24 June 2007, we googled ‘shut the X up’ (with the quotation marks) and ‘beat the X out of’ (again, with quotation marks), where for X we substituted {hell/fuck/shit/bejesus}. Table 5 shows the number of hits for each taboo term in the two constructions and the percentage of the total number of occurrences of each construction we googled.
Table 5 Occurrences on 24 June 2007 of hell, fuck, shit, and bejesus in shut the X up and beat the X out of
Certainly, the bejesus, being somewhat archaic, should not be expected to occur often. But it did occur in 1.3% of the tokens of the B-construction phrase we tested (beat the X out of), whereas it occurred in only 0.000024% of the tokens of the G-construction phrase (shut the X up). So, at least with respect to these two phrases, it is highly disfavored in the G-construction, to the point where we could say that the five people who used it are exhibiting an idiosyncracy. We must conclude that all other taboo terms tested are grammatical in both constructions for these phrases. However, the shit is clearly much less common in the G-construction, accounting for only 1% of the hits, and the fuck is much more common for the G-construction, accounting for a whopping 69.6% of the hits. Still, the {hell/fuck} occur with high frequency in both constructions.
With this in mind, it is instructive to return now to our original corpus and compare the two constructions with respect to the {hell/fuck}, adding in the taboo term the heck, since it occurred frequently in both constructions. Table 6 restricts the data from table 4 to just these items. We see that the fuck shows a strong preference for G over B, that the hell has a weaker preference in the same direction and that heck favors the B-construction. Given the intuitive ranking fuck>hell>heck on a scale of strength or rudeness, this suggests that the G-construction is ruder than the B-construction. We believe that this is indeed the case, and that it has to do with another difference between the two constructions, to be dealt with in the next subsection.
Table 6 Restricted to fuck, hell, and heck
Based on table 6, the choice of expletives in the G-construction provides us now with a criterion to establish whether shut the hell up or leave her the fuck alone should be regarded as instances of the G-construction, in spite of the fact that they do not involve motion. We simply note that the particle verbs shut up, leave alone and others readily combine with the three expletives that show up in the G-construction, but not nearly so readily with the other ones. We already compared the frequent shut the {hell/fuck} up with the very infrequent (1%) shut the shit up and the almost nonexistent shut the bejesus up in table 5. A quick google search on 24 June 2007 of ‘leave her the X alone’ (with quotation marks), substituting hell, fuck, shit, crap, and bejesus for X resulted in a total of 5169 hits, of which 4610 were for fuck (89.2%), 554 for hell (1.1%), 3 for crap, 2 for shit, and none for bejesus.
4.5. Directiveness
Many occurrences of the G-construction are directive. Most common among these are commands, such as Get the hell off my property!, but adhortative sentences also frequently occur, such as Let's get the hell out of here! and We should get the hell out of this place. A third type of clause, which we have treated as an embedded directive, is the complement of a predicate such as order or tell: The sarge told us to get the hell out. On the other hand, sentences such as So we got the hell out or They beat the hell out of him are nondirective, regular descriptions. We marked up all our data for directiveness, and found a very strong correlation between the G-construction and directiveness, shown in table 7.Footnote 8 While only a handful of the occurrences of the B-construction were directive, 75% of the occurrence of the G-construction were.
Table 7 Directiveness for two constructions
It seems natural to assume that it is more at variance with common rules of politeness to tell someone to get the hell out than it is to report that someone got the hell out. Hence the ruder terms are best saved for directives, whereas the more socially acceptable ones, such as heck, will be relatively more common in nondirective utterances. Of course, we do not claim that commands involving the G-construction are always rude or that they must convey something negative to the hearer, but rather that they frequently are, and do.
With the B-construction, the force of the expletive is simply emphasis. When an editor tells an author ‘We're going to sell the hell out of this book’, nothing nasty is implied. Quite generally, the expletive is not related to the attitude of the speaker toward the listener. With the G-construction, however, the attitude of the speaker to the listener does appear to be relevant, at least with directives. Directive utterances (with or without expletives) often convey bluntness, rudeness, or anger. These connotations are so strong that textbooks for second language learners of English include warnings against them in their sections on pragmatic competence (Bland Reference Bland1996 offers a thorough discussion). Of course, both the situation of the speech act and the intonation of the speaker can counter this effect, in which case the expletive is simply an indication of urgency and/or speed: Jones got the hell out of the burning barn.
Indirect (embedded) directives, usually with speech act predicates such as tell, order, give orders, etc., derive their properties from direct speech acts. So they too express some kind of emotional connotation, ascribable, in this case, not to the speaker of the entire utterance, but to the agent of the speech act verb, who is held responsible for the content of the embedded clause.
4.6. Choice of coda
The syntactic category of the coda, i.e. the expression following the expletive, may also serve to distinguish the G- and B-constructions.
The various types of codas
Table 8 lists the type of codas in our data for the two constructions at hand. While both constructions show mostly PPs in coda position, the G-construction also has codas of the category P (prepositional particle), Adverb or Adjective.Footnote 9 This is enough, of course, to show that G and B differ in this regard as well. The differences are, moreover, not merely statistical in nature, but categorical, as predicted by our account. We have argued that the B-construction involves a verb alternation which turns a direct object into a PP complement headed by out of. Hence it is to be expected that only PP codas are to be found. Even the closely related particle codas do not show up in the B-construction, because they cannot host the verb's theme argument.
Table 8 Codas for two constructions
We find the status of codas interesting enough to warrant additional discussion in the following subsections.
The coda is a PP
It is clear from table 8 that PPs are by far the most common type of coda. Yet it is equally obvious that not all PPs make good codas for our two expletive constructions. Here we address the criteria that are relevant to PP selection.
First, we note that the B-construction always involves out of or its phonological variant outta. Sentences such as (Reference Lee39), from our corpus, are not problematic for this claim since they are not examples of the B-construction:
(39) That church attempts to scare the bejesus into people.
The bejesus here is not telling the degree to which people will be scared, but is, rather, the theme argument, while people is the locatum argument. That is, the reading of (39) is that the church uses scare tactics to make people believe in their teachings. In fact, we can substitute other theme arguments with no problem:
(40) His dad tried to scare some {sense/caution/decency} into him.
However, a quick google search of ‘scare the X into’ on 24 June 2007, revealed that some people do use scare the X into as a variant of scare the X out of when X is a taboo term. We found only one hit for scare the fuck into, but 610 hits for scare the shit into (as compared to 132,000 hits for scare the shit out of). We also found 32 hits of scare the bejesus into, but most (if not all) of them, like our (39) above, were in religious contexts and were to be read similarly to (Reference Lee39).
While there is nothing more to be said about the PPs in the B-construction, there is much to say about them in the G-construction. In table 9, we give an overview of the Ps found in our data set for the G-construction.
Table 9 Prepositions in the G-construction
The most common preposition is the complex P out of, but others are attested as well, in particular, directional prepositions indicating motion away from some pragmatically-given pivot, usually the speaker or the subject of the sentence. For prepositions that are polysemous (the majority), directional uses may be acceptable whereas nondirectional ones are not to the same degree. For example, we found (41a) on the Internet, but no one we asked accepts (41b):
- (41)
(a) Run the hell about the house.
(b) *Let's talk the hell about the war.
Some fairly idiomatic cases do exist, however, such as this one from the Internet:
(42) Pay attention and get the hell with it.
In short, it is the predominance of PPs with out of that makes the G- and B-constructions look similar at first, and conversely it is the possibility of a range of Ps that sets the G-construction apart from the B-construction.
The coda is a particle
When the coda is a particle, we are dealing with the G-construction. The B-construction only accepts out of-PPs as its coda (with the minor exception we noted in section 4.6.2). Particle verbs are common in the G-construction, however, especially shut up, hurry up, get out, get off, get up. Less common were back up, chill out, drink up, drive off, freak out, go out, lock up.
The coda is an adverb
Adverbs in coda-position form a very small group of items with clear PP-like characteristics: away, back, back down, down, here, there, home, inside, outside and now. The cases with now and here in our data set are special in that they do not involve a verb (for example: Right the hell now!), and so they do not immediately strike one as candidates for either the G- or the B-construction. We defer the discussion of these cases to section 4.7, where we will consider a number of other problems as well.
The items back, down, home, inside and outside are all adverbs that behave like directional PPs in some ways, and like particles of particle verbs in other ways. For instance, they may be conjoined with directional PPs:
- (43)
(a) The prospectors went back and into the cave.
(b) The police went inside and into the kitchen.
Jackendoff (Reference Jackendoff, Stephen and Paul1973) mentions similarities between back, down, etc. and PPs, such as the fact that both may be the locative argument of put:
- (44)
(a) Sheila put the clothes {in the closet/inside/on}
(b) Irving put the books {on the shelf/there/away}
Another test he uses for PP status is modification by right:
- (45)
(a) The eager dwarves rushed right {into her arms/inside/back}.
(b) Just go right {to the traffic light/back/in/home}.
On the other hand, these adverbs may also be found in preverbal position in verbal compounds, like particles but unlike full PPs: downsizing, homecoming, inside-trading are like uplifting, outgoing, ongoing but unlike *inthebackstabbing, *at-home-cooking, etc. It is conceivable that the sheer bulk of full PPs is what makes them resistant to incorporation.
Another difference, already noted by Jackendoff, concerns the positional possibilities of particles, in contrast to PPs:
(46) Sheila put {on/back/*in the closet} the clothes.
While we generally agree with the asterisk in (46), we note that with enough context and appropriate prosody, the in the closet variant can be made to sound relatively natural. So we believe that here, as well as in the case of incorporation, heaviness may come into play. Light elements may appear in between the verb and the direct object, heavy items are disfavored, unless the direct object itself is heavy:
(47) Sheila put in the closet all the clothes that she had found on the floor.
The best argument for the view that heaviness is at stake comes from the well-known fact that pronouns do not normally appear after particles, unless they are given extra stress (indicated here by boldface) or made heavier by attached modifiers:
(48) Sheila called up {*him/him/only him}.
Given the similarities among PPs, particles, and adverbs of the back-, down-, here-class, we may view the codas of the G-construction as PP-like entities.
The coda is an adjective
Among the remaining expressions, primarily adjectives, the most prominent by far is alone, in the combination leave alone. This adjective exhibits PP-like behavior. Alone, together with some other adjectives beginning in a-, such as asleep, awake, alive, afraid, ajar, tends not to occur in the usual adnominal position of attributive adjectives (Markus Reference Markus and Antoinette1998: 135):
(49) ?an {alone/afraid/alive} student
although, as Markus notes (following Quirk, Greenbaum, Leech, & Svartvik Reference Quirk, Greenbaum, Leech and Svartvik1985), the results are generally better when the adjectives in question are modified:
- (50)
(a) a somewhat {alone/afraid} student
(b) a really alive student
Historically, most a-adjectives derive from PPs, and the positional preferences of these adjectives still reflect their origin (Markus Reference Markus and Antoinette1998). Thus the OED, s.v. a-, states that akin derives from of kin, with a- as a reduced form of of (compare also outta<out of, e.g. we got to get outta here), and that ado stems from at do, while yet others, e.g. abed, ashore, alive derive from on: on bed, on shore, on live. While alone does not actually derive from a PP, but from the combination all-one, it was most likely reanalyzed at some point as a member of the class of a-adjectives. The existence of the adjective lone, without the a-, provides some evidence for such a reanalysis. Further evidence can be seen in the behavior of degree modifiers. Like PPs, but unlike most adjectives, a-adjectives do not use the degree modifier very but rather the complex modifier very much:Footnote 10
(51) very (*much) {nice/talented}
- (2)
(a) *very {out of her mind/into Greek art}
(b) very much {out of her mind/into Greek art}
- (52)
(a) ?very alone
(b) *very aghast
(c) very much {alone/aghast}
In other words, it looks like the distributional properties of a-initial adjectives coming historically from a PP source were grammaticalized, spreading in turn to alone on the basis of phonological similarity.
However, while we believe that the PP-like characteristics of alone may have helped the combination leave alone in becoming by far the most palatable host for the hell and its ilk, it is not in fact the only such combination that is attested. Other examples of the form verb+expletive+adjective are given here:
- (53)
(a) … let me get back to work and then let me the hell alone. (Lee Reference Lee1948: 101)
(b) … if I shoot it, it goddamned well better drop the hell dead!
(c) Get the hell better so we can all get shit faced.
(d) … and sometimes you just want to get the hell drunk.
(e) Be sure to get the hell stuffed!
The examples all appear to involve complex predicates: leave alone, let alone, drop dead, get better, get drunk, get stuffed, stay clear of, steer clear of. Adjectives occurring with the more common verbs be or become are hardly attested.
4.7. Special cases
We find the hell, the fuck and other taboo terms following some intensifiers:
- (54)
(a) So the hell off the mark.
(b) Right the hell {now/here}.
(c) All the hell around the world.
(d) Way the fuck over in Canada.
The intensifier licenses the taboo terms; its absence yields ungrammaticality:
- (55)
(a) John's in *(way) the fuck over his head.
(b) Just watch; she'll be *(all) the fuck over him at the party.
(c) She dove *(deep) the fuck under – I mean deep, baby.
The examples in (55)–(56) are neither B-constructions nor G-constructions. Rather, this use of taboo terms can appear anywhere the appropriate intensifiers can appear. In (56) they go with intensifiers of PPs. But in (57) they go with intensifiers of an AP, a VP, and an NP:
- (56)
(a) AP: You went so the fuck wrong, I can't begin to explain.
(b) VP: She so the fuck doesn't understand.
(c) NP: He's so the fuck a liar.
Just as we saw with wh-expressions in (15) above, which also license taboo terms, the addition of the hell or the fuck adds force, so much force, in fact, that the fuck is more common than the hell in this usage, and degree adverbs that do not have a strong intensifier sense cannot license them:
(57) *She's {a little/a bit/somewhat} the fuck frightened.
The intensifier that most commonly licenses taboo terms is so. Indeed, the frequency of taboo terms after intensifier so may be responsible for the appearance of them after other uses of so. For example, the affirmative pro-predicate so, which is anaphoric rather than an intensifier, can license taboo terms:
- (58)
(a) A. I'm tired of all this fighting.
(b) B. Oh, yeah? Well, so the fuck am I!
And the idiom so what can be broken up by inserting a taboo term:
(59) So the hell what?
The licensing of taboo terms by intensifiers is a relatively new phenomenon in English. Yet it should be no surprise. We already have taboo terms functioning as intensifiers with wh-words and in the B- and G-constructions (the history of this development is presented in section 5). It is a quick jump, then, to taboo terms combining with other intensifiers to make a kind of super-intensifier, so that so stupid becomes so the fuck stupid and too dumb to understand becomes too the fuck dumb to understand. In the judgments of our Swarthmore informants, the expletives used with intensifiers are those we find with the G-construction, in particular the hell/fuck, rather than the more mild ones that are found so often in the B-construction. Compare:
- (60)
(a) so the {hell/fuck} stupid
(b) *so the {shit/crap} stupid
5. History of the two constructions
The B- and G-constructions first appear in printed texts toward the end of the nineteenth century and these early attestations of both constructions involve religious terms only, such as the devil and the hell. Accordingly, we conjecture that they originated around that time. It is conceivable that they arose much earlier in the spoken language, but we have no evidence that this was actually the case. It is always a difficult task to estimate when a new word or idiom enters the language on the basis of written language, and this difficulty is exaggerated when dealing with taboo terms, which one might expect to be limited to the most informal of exchanges. However, some of the taboo terms were obviously not too rude for the Victorians to use in print, given that expressions like who the devil or why the deuce show up quite regularly in 19th-century novels, as they did before and after. If the use of taboo terms in the G- and B-constructions does not show up much in 19th-century sources, we think it likely that this is not because it was considered insufferably crude, but because the constructions were too new and not yet in wide circulation. Evidence for this claim comes from one of the first attested cases of the G-construction that we have been able to find, in a Senate report:
(61) Get the hell out of here. (Lexow & Cantor Reference Lexow and Cantor1895: 4665)
Proceedings of US Senate Hearings often contain vivid, informal language, but certainly not anything considered unprintable. If it were deemed extremely rude, we would expect rather different sources for the earliest occurrences of the hell in the G-construction, such as personal correspondence or naturalistic novels.
An early occurrence of the B-construction is:
(62) Say, old man, you must have knocked the devil out of her coming over alone in twenty-nine days! (Slocum Reference Slocum1900: 47)
The context of this example is relevant. The pronoun her is referring to a sailing vessel, and so it is clear that the use of knock the devil out is idiomatic, and does not refer to some exorcism ritual. Earlier occurrences of the devil+out typically denote religious practices, and do not appear to be idiomatic.
5.1. The B-construction
It is likely that one origin of the B-construction is exorcism: beating the devil out of somebody. Most early occurrences of beat the devil out of X describe either actual exorcism of the devil or a beating meant to rectify a rebellious or wayward spirit. The origin of the B-construction is hard to date because of the difficulty in distinguishing the earlier exorcism interpretation from the later intensifier reading. Sometimes the choice of the verb may help here. If it is drive the devil out of X, or cast the devil out of Y, we assume that exorcism is the intended interpretation. With many modern examples, however, we can be unequivocal in our analysis of them as examples of the B-construction because no exorcism reading is possible; often they do not even involve verbs of physical or mental abuse, and many employ the taboo term the fuck, which precludes an exorcism reading.
When the verb is beat or scare, matters are a bit more complex. Our best guess is that the B-construction goes back at least to the latter half of the 19th century. The oldest occurrence of the B-construction we have been able to find in Google Books dates from 1895:
(63) Yes, Loubitza will beat the devil out of her when she gets her home – her and her broken jar! (Daintrey Reference Daintrey1885: 74)
At some point, the expletive must have ‘bleached’ to the extent that it became solely an intensifier. The original meaning of exorcism still shines through in newer variants such as beat satan out of or the one in the following example from the Internet, where the context shows that religious and intensifier uses are being played upon:
(64) When ‘Charlie's Angels’ beat the Beelzebub out of Adam Sandler's ‘Little Nicky’, bringing the second week gross of Drew Barrymore, Cameron Diaz, and Lucy Liu's T&A blockbuster to a heavenly $75 million, the triumph of seraphim over Satan seemed divine. And in true Hollywood tradition, the next question wasn't ‘How do I thank thee, Lord?' but rather ‘How fast can we make a sequel?’ For Sony, it's not going to be a piece of angel food cake.
The substitution of the hell for the devil in the B-construction may well have taken place under the influence of an earlier change in English strong language involving what the devil, etc. being replaced by what the hell. This change probably started in the late 18th or early 19th century, well before the earliest attestations of beat the hell out of X, etc. The oldest attestations we have been able to find are from the early 19th century, e.g.:
- (65)
(a) I wonder what the hell brings us here again? (Cunningham Reference Cunningham1826: 146)
(b) Him! why what the hell is he but a priest? (Carleton Reference Carleton1845: 244)
The substitution the devil>the hell may also help explain why hell is preceded by an article, something it normally does not have in English (cf. Go to hell, Hell is a hot place, etc.), where hell is treated, like heaven for that matter, as a proper name. We should note, however, that up to roughly the 1920s, the more common variant in the B-construction is beat/punch/etc. hell out of X, without the article. Compare e.g. the following examples:
- (66)
(a) Sheridan […] used some rather strong language and said that he would ‘knock hell out of Stuart if he could get at him’. (Indiana County Gazette, November 6, 1890)
(b) Tom Tarkington also testified that Goodman told him during the day that he was going to whip hell out of the appellant. (Court of Criminal Appeals of Texas, June 12, 1901)
(c) One of them mots that do be in the packets of fags Stoer smokes that his old fellow welted hell out of him for one time he found out (Joyce Reference Joyce1922: 247)
The addition of the article was probably reinforced by the earlier adoption of the hell in wh-questions, where the bare noun is very rare and appears to be possible only when a preposition is present:
- (67)
(a) What *(the) hell are you doing?
(b) What in hell are you driving at?
Even more striking is the later occurrence of the fuck, again obtained by simply substituting one noun for another, leaving intact the article. Had it not been for the prior existence of the hell and the devil, we would not have expected the fuck to be used in this way. (The same argument applies to rarer forms such as who in the hell, who in the world which led to the peculiar substitution form who in the fuck.)
Apart from religious contexts involving exorcism, there appears to be another source of the construction, not involving expletives such as the devil but, instead, the life: beat the life out of X, scare the life out of Y, strangle the life out of Z, etc. These involve a similar bleaching, moving from descriptions of attempts on someone's life to mere intensification:
(68) And that's what's frightening the life out of me. (Christie Reference Christie1940: 156)
We note that words for life and death are often used for purposes of intensification, compare e.g. These sermons bore me to death or I can't understand for the life of me what the hell he is talking about, and the use of dead as an adverb of degree: dead wrong, dead tired, dead certain. Again, it is difficult to pinpoint exactly when the literal use developed into the intensifier use, but it is plausible that this happened in the 19th century.
Attested from the second half of the 19th century onward is beat/scare the living daylights out of X, cf.:
(69) … he started to club the living daylights out of the beast with his gun (Decatur Morning Review, 10 September 1890)
It is hard to tell whether the scatological variants beat the shit out of X or scare the crap out of Y are more recent than the religious ones, because of the strong taboo which the Victorian age put on words for excrement. However, the Oxford English Dictionary (2nd edition, Simpson & Weiner Reference Simpson and Weiner1989) has not come up with anything antedating Leonard Cohen's 1966 LP Beautiful Losers, and neither have we. It is highly likely, then, that the expression is not as old as the variants involving the devil and his residence.
The use of the fuck in the G-construction is probably from around the same time. The oldest attestations in print we were able to find are from the late 1960s:
- (70)
(a) Get the fuck off me! (Rollins Reference Rollins1967: 113)
(b) Get the fuck out of here. (Walker Reference Walker1968: 225)
At the moment, we see a flowering of ever-longer variants on the Internet:
- (71)
(a) … get the living fuck out of my house!
(b) And then I get the proverbial fuck out.
(c) You people are mother fuckers that deserve to get the everliving fuck beaten out of you for being worthless human beings.
(d) He squeezed off a couple of rounds in the air and we thought that would be a good time to do a three-point turn and get the raging fuck out of there.
(e) He got into his car yelling something at her and she said ‘That's right you effing abuser, get the effing hell out of here’.
(f) … get the flying fuck out of my face or I'll puke on you
Similar developments can be noted in the B-construction, as well as wh-intensification. For the latter, compare such examples as:
- (72)
(a) … then where in the holy living fuck did all those cavemen and cannibals and headhunters and Mesolithic horticulturalists come from?
(b) How the fickety fuck-bums are you ma wee sex-communist?
(c) … how in the blue fuck is that even possible???
(d) … what the friggin hell is a ‘bisexual’?
(e) Who the motherfucking fuck do you think you are?
5.2. The G-construction
The G-construction is much easier to identify, even in its first occurrences, than the B-construction. The oldest attestation we have been able to find, so far, is (62) above, repeated below, from 1895:
(73) Get the hell out of here.
It appears, then, that the B- and G-constructions are about equally old. There is no strong evidence that one gave rise to the other, but the similarities between the two constructions, and the fact that they both at some point substituted the fuck for the hell, suggest that they are closely related and developed in tandem. If one construction derived from the other, we surmise that the G-construction was derived from the B-construction. This might explain why the oldest attestations of the G-construction involve the preposition out of and why out of-PPs are still by far the most common type of coda. The precise mechanisms and manner of such a derivation, if correct, remain murky, however.
6. Summary and outlook
The B- and G-constructions studied in this paper seem at first glance to be a single construction, but we have shown that they differ in both syntactic and semantic structure. In the B-construction the taboo term has the grammatical role of direct object, but is not an argument of the verb, the thematic argument being the object of the complex preposition out of. Non-thematic dummy objects are precluded. In the G-construction there is neither a direct object nor a thematic argument of the verb. The B-construction employs only a PP headed by out of, while the G-construction most frequently employs out-of PPs, but also allows a variety of adjuncts, including other PPs, particle Ps, Adverbs, and Adjectives. The verbs in the B-construction are transitive and typically have to do with physical or mental abuse, while the verbs in the G-construction are mostly intransitive. There is a wide range of taboo terms in the B-construction, but the G-construction tends toward only a few and the strongest. This last fact correlates with the fact that the G-construction is frequently a directive.
In both constructions the taboo term functions as an intensifier, of the action of the verb in the B-construction and of the attitude of the speaker in the G-construction. Other positions in which taboo terms function as intensifiers include after wh-words, after degree intensifiers such as so, and in emphatic denial or assertive statements. Our analysis here suggests that the use of taboo terms as intensifiers spread from the wh-constructions to the B- and G-constructions and, finally, to the degree intensifier constructions.
The occurrences of taboo terms in a great many different constructions makes sense only if we assume two things:
• some constructions have a special emphatic status in the language, and show a tendency to employ taboo terms to express that status
• once items are enlisted by some constructions, they may (but do not have to) spread to other constructions as well
The spread from one construction to another might be viewed as a type of grammaticalization. The historical developments we have discussed in the two major constructions analyzed in this paper strongly indicate that pragmatic cohesion (evidenced in the fact that these constructions have a special emphatic status) suffices to initiate such grammaticalization, which can then have a ripple effect on historical change, so that we find similar taboo terms functioning in similar ways in structures that are disparate both semantically and syntactically.
Theories of grammar that assume constructions as their basic building blocks, such as Construction Grammar (Fillmore, Kay & O'Connor Reference Fillmore, Kay and O'Connor1988, Goldberg Reference Goldberg1995), to some extent also HPSG and TAG-grammar, predict that phenomena such as the one described in this paper might readily occur. On the other hand, theories which do not admit constructions into their repertory of syntactic primitives, such as Chomsky's Minimalist Program (Chomsky Reference Chomsky1995) or classical versions of categorial grammar (Lambek Reference Lambek1958), do not predict their occurrence. Rather, in such theories the appearance of these particular taboo terms in all these different constructions is mere coincidence, and leads to no predictions about the possible syntactic environments that new taboo terms might occur in. For example, if a new term were coined for a sex act, we might expect it sooner or later to appear in a subset of the syntactic positions that fuck can occupy in theories that admit constructions as syntactic primitives, but not in theories that don't. Indeed, fictive expletives do occur in the relevant syntactic slots. The failure of the latter set of theories to make this kind of prediction is problematic, given that in censored environments people will even substitute nonsense words such as ‘bleep’ in the relevant syntactic slots.
