Version classiqueVersion mobile
OpenEdition Books

La Négation

Stéphanie Bonnefille
Sébastien Salbayre

Linguistique : négation, stratégies discursives et processus mentaux

Negation in the margin of grammar

Bert Cappelle


On a linguistically naive view of sentence negation, all positive sentences have a direct negative counterpart and all negative sentences have a direct positive counterpart: after all, negation is just a matter of changing the polarity by means of an element like not. In this paper it is shown that this view is, indeed, all too simplistic. Three marginal sentence types in English are discussed: one which is always positive and cannot contain negation (unless under special conditions), another which is always negative and cannot be made positive by removing the negative items, and yet another where adding or removing a negative item does not seem to cause any change in meaning

Texte intégral

1This paper has a look at the category of negation from what might be considered as the periphery of grammar. In this periphery, we find patterns which are less general or well-known than constructions like the passive or the imperative, but which are no less part of the grammatical competence of speakers. The marginal patterns discussed here are already special because of certain idiosyncratic syntactic and semantic properties, and they are furthermore interesting with respect to negation.

2In section 1, I will first introduce the form and concept of negation by means of a series of stereotypically negative clauses: detention lines written on a blackboard (I will not...). A discussion of them will lead me to venture a general definition of negation. I will then propose three simplistic views of negation, which in the subsequent sections will be argued to be impossible to be upheld.

3Section 2 is devoted to the correlative comparative construction (the X-er, the Y-er). I will show that negation of the second part of this pattern is only possible under certain conditions, thus rejecting the view that adding a negator is an easy operation that can be applied to any positive clause.

4In section 3, I will more briefly address conditionals of the No X, no Y type. This pattern proves wrong the view that negation is an aspect of meaning that is added to pre-existing propositions, and that all negative sentence types must therefore have a corresponding positive version.

5Section 4 deals with the so-called constant polarity tag (as in So, you think you're smart, do you?). I will argue that this term is somewhat unfortunate, since we can observe that a similar antagonistic meaning tends to be expressed by a reversed polarity tag after a negative anchor (e.g. Oh, but you wouldn't care, would you?). English seems to be in the process of developing a rule by which antagonistic tags keep their positive polarity constant, irrespective of the polarity of their anchor. This kind of tag, then, shows that there are constructions that do not allow to be made negative, again rejecting the more simplistic view to the contrary.

6The main findings are summed up in section 5.

7Much of the linguistic material on which this study is based is drawn from The Simpsons, a series known for using language that resides in the margin of grammar and in the margin of what is grammatical.


Well if you're so sure what it ain't, how about telling us what it am! (Moe Sizlack in The Simpsons, episode “Lisa the Skeptic”)

1.1. Starting point: Blackboard lines from The Simpsons

  • 1 For a full and annotated list of blackboard quotes, see The Simpsons Archive on (...)

8The Simpsons, the longest-running and arguably one of the wittiest animation series in TV history, has a standard opening sequence in which we see the ten-year-old Bart Simpson writing lines on a blackboard during after-school detention. The lines that the popular cartoon character has been forced to write differ from episode to episode, but they are almost invariably negative sentences. A small sample of these blackboard opening lines is given below:1

(1) a. I will not waste chalk
b. I will not use abbrev.
c. I will not mock mrs. dumbface
d. Funny noises are not funny
e. Pork is not a verb
f. Grammar is not a waste of time
g. Cursive writing does not mean what i think it does
h. High explosives and school don' t mix
i. There was no roman god named “fartacus”
j. I saw nothing unusual in the teachers ' lounge
k. I won't not use no double negatives

9Taken together, these sentences provide us with an excellent overview of standard negative clause formation in English. As is well known, negation is typically marked by the word not. This so-called analytic negator can be added directly to an auxiliary verb, as in (lac), or to an inflected form of the copular verb be, as in (1d-f). Lexical verbs like mean or mix, by contrast, are not followed directly by not but need do-support, as is illustrated in (1g-h). Sentences (li) and (1j) show that negation need not be marked on the verb, but sentences with non-verbal negation can be replaced, without a change of meaning, by sentences in which the negation does appear on the verb (There wasn't a Roman god named “Fartacus”; I didn't see anything unusual in the teachers'lounge). Sentence (1k) illustrates a phenomenon that is not considered Standard English: single semantic negation with multiple grammatical markings.

10Spelling out what it is about the blackboard quotes in (1) that makes them funny may reveal something crucial about the semantics of negation. I see at least three reasons for why they are (mildly) humorous.

11First, some of them contain a mismatch between their form and their content, thereby self-referentially denying the proposition they express. This is obviously the case in (lb, c, f, and k)—although in (lk) Bart is not really contradicting his own statement, since he is using a triple and not a double negation (which makes it all the worse, of course). If we transcend the level of the sentence itself and look at the entire discourse situation, with Bart Simpson writing the same line all over the blackboard, then it is clear that (la), I will not waste chalk, also hinges on this humorous form-content mismatch.

12Second, the fact that we only witness Bart's punishment means that we have to mentally reconstruct what his mischief or misbehavior has been. The blackboard line is the only clue in this reconstruction process. It is this suggestiveness that makes the line funny. Humor often resides in what is not said or shown rather than in what is said or shown—indeed, a joke is always more funny than its explanation. (Accordingly, I do not expect the reader to chuckle at these paragraphs.) So, in the case of Funny noises are not funny, for example, we are invited to picture to ourselves a class-room situation in which Bart found, to the teacher's annoyance, that funny noises were funny. This particular blackboard line is also a slight play on words: funny can have negative and positive connotations: “odd, strange” and “comical, humorous.” The point the teacher wants to drive home is that just because a noise is funny in the one sense does not mean it is funny in the other sense; in fact, as far as the teacher is concerned, if the noise is funny in the bad sense, it is, ipso facto, not funny in the good sense. In other words, we are led to understand that Funny noises are funny is, rather surprisingly, not a tautology, if you think about it. The blackboard line Funny noises are not funny therefore functions as a sort of punch line to a one-line joke. This is also the case for Cursive writing does not mean what I think it does.

13Third, and perhaps most relevant to grasp how negation really functions cognitively, the line I saw nothing unusual in the teachers' lounge makes us laugh because we know that Bart's punishment has the reverse effect of what is intended. By making Bart deny that he saw something unusual, the teacher only prods everyone's imagination to speculate about what Bart could possibly have seen. The proposition that is negated here is strongly felt to be closer to the truth than its denial. In general, negating an insinuation merely reinforces it—a wisdom Nixon was unfortunate enough to gain after he proclaimed I am not a crook on the national television during the Watergate scandal. The fact that negative sentences inevitably evoke their positive counterpart has been known for a long time by both psychologists and linguists. Their favorite example is Don't think of an elephant!, which curiously makes the mind rush to think of an elephant. (See Lakoff, 2004 and Sweetser, 2004, for some recent applications of this cognitive mechanism.) It is as if listeners mentally delete the negator when they hear a negative sentence.

1.2. A rough definition of negation

  • 2 Negation should be seen here as shorthand for ordinary clause negation. I am not concerned with ne (...)

14Having had a look at some aspects of form and meaning of negation, we are now at a point where we can say what negation is. Not to put too fine a point on it, we might define negation as follows:2

(2) Negation is a grammatical operation, typically effectuated by not, that reverses the polarity of an implicitly understood background clause, thus producing a clause that expresses a proposition which is semantically the opposite of the proposition expressed by the background clause.

1.3. Aim of this study

15Proposing a definition of negation is not the central aim of this paper, though. On the contrary, I want to say what negation is not—but not without at least trying to state what negation is, as I did in section 1.2 above.

16In what follows, I will negate three interrelated claims that we might, rather simplistically, consider as received opinions about negation:

(3) a. Negation is a simple operation.
b. All positive clauses have a negative counterpart.
c. All negative clauses have a positive counterpart.

17I will show that the propositions in (3) are false by discussing negation in three peripheral constructional patterns in English. My commitment to dealing with marginal patterns is in line with research in Construction Grammar (see Fried and Östman, 2004, for a thorough but accessible introduction to this theoretical framework). The philosophy behind this approach is that we miss out on crucial facts about language if we restrict our attention to core linguistic phenomena. Thus, if we only looked at negation in free-standing, simple declarative clauses like the ones in (1), we would end up with misguided views of how negation relates to clause structure.


The older they get, the cuter they ain't
(Aunt Patty in The Simpsons, episode “Lisa' s First Word”)

2.1. A brief introduction to the correlative comparative construction and to its meaning

18The sentences in (4a-c) below illustrate a pattern that is variously known as the “correlative comparative construction” (Huddleston, 2002b, 1135-1137), the “comparative correlative” (Culicover and Jackendoff, 1999; Borsley, 2004a, 2004b; Den Dikken, 2003), the “comparative conditional (construction)” (McCawley, 1988a; Michaelis, 1992; Beck, 1997; Declerck and Reed, 2001), the “double comparative construction” (Thiersch, 1982), or the “BCHF construction” (standing for the idiomatic instance “the bigger they come, the harder they fall,” Fillmore, 1986):

(4) a. The older they get, the cuter they are.
b. The more vegetables I eat, the worse my gastric problems get.
c. The harder we try, the less we achieve.

  • 3 As is customary in linguistics, curly brackets (“{}”) enclose linguistic alternatives, separated b (...)

19This pattern evokes paired scales, for example age and cuteness, and conveys the idea that movement along the one scale correlates with proportional movement along the other. The correlation need not be positive: (4c) refers to an inverse correlation between change on an effort scale and change on a success scale. The construction can be closely paraphrased by means of as: “As they get older, they get correspondingly cuter.” A conditional interpretation also often holds (e.g. “If they get older, they get cuter”), but as Declerck and Reed (2001, 28) point out, “A sentence like The longer I knew him, the less I understood him is not interpreted as” If I knew him longer, I understood him less. “There is no conditional interpretation either in The more I listened to him yesterday, the less I could believe him.” Moreover, Culicover and Jackendoff (1999, 545) show that the construction does not allow a counterfactual instantiation (e.g. * The older they had gotten, the cuter they would have been), which also makes it closer to an as-comparative than to an if-conditional (cp. {* As/If} they had gotten older, they would have been (correspondingly) cuter).3

2.2. A closer look at the syntax of the correlative comparative construction

20The correlative comparative construction is obviously a noncore pattern in English. Nevertheless, it has close parallels in many other languages, such as Dutch, German, French, Polish, Russian, Hungarian, Latin, Greek, Hindi, and Mandarin Chinese (McCawley, 1988a; Michaelis, 1992; Savelli, 1993; Den Dikken, 2003; Borsley, 2004a).

21Before we turn to a discussion of negation in this construction, let us briefly present some of its general structural facts, as far as English is concerned.

22First, although the two asyndetically linked clauses in this construction may superficially look completely equal to each other, there is evidence that the second clause is the main clause and that the first clause is a subordinate clause. For one thing, if we add a question tag to the construction, we find that it is the second clause that determines the form of the tag, not the first clause—see (5a). For another, if we embed the construction as a whole in a subjunctive that-clause, we find that the subjunctive mood can be marked on the verb of the second clause, not on the verb of the first clause—see (5b).

(5) a. The older they get, the cuter they are,{* don” t/aren’t} they?
b. It's vital that the older Harry {* get/gets}, the cuter he {be/is}.

23Second, there is a syntactically more basic but less common version in which the main clause precedes the subordinate clause, and in which the comparative phrase in the main clause appears in its canonical position and is not generally introduced by the:

(6) a. They are (% the) cuter the older they get.
b. My gastric problems get (% the) worse the more vegetables I eat.
c. We achieve (% the) less the harder we try.

24We will come back to this related pattern. Let us call it henceforth the “head-clause first pattern.”

25Third, under certain conditions, the clauses can or have to be reduced. The copula be can be omitted if the subject has a generic, non-specific interpretation. In any case, the copula be cannot be omitted if the subject is a pronoun—compare (7a) with (7b). It is also possible for the second clause to be reduced to just the comparative phrase. Moreover, if the first clause is reduced this way, so must be the second—compare (7c) with (7d). The comparative phrase can itself also be reduced, as in the conventionalized expression (7e), which may be analyzed as elliptical for “The more haste (there is), the less speed (you will gain).”

(7) a. The higher the buildings (are), the lower the morals (are).
b. The higher the buildings (are), the more vulnerable they *(are).
c. The higher the buildings (are), the more vulnerable (they are).
d. The higher, the more vulnerable (* they are).
e. More haste, less speed.

26Fourth, for some speakers, the comparative phrase can be followed by that:

(8) The older (% that) they get, the cuter (% that) they are.

27Fifth, subject-auxiliary inversion is occasionally found in the second clause:

(9) a. The more he studies, the lower is his motivation.
b. The more he studies, the more do words deceive him.

28Finally, embedding is possible, as we have seen in (5b), but not if it involves so-called wh-extraction:

(10) a. I agree that [the older they get, the cuter they are].
b. * There are some kids who [the older they get, the cuter are].

29Let us now have a look at how negation interacts with this construction.

2.3. The use of negation in the correlative comparative construction

2.3.1. An example of what is not (very) grammatical

30In one particular episode of The Simpsons, Marge Simpson, mother of three children, is bringing up memories of when Lisa, the second child, was born: Everyone agreed that Lisa was a beautiful baby. We then get a flashback to that time, with Marge's older chain-smoking twin sisters, Patty and Selma (referred to as the gruesome twosome by Marge's husband Homer) dropping by to see little Lisa. Bart, Lisa's older brother, cannot stand all the attention that is lavished on this newcomer, and tries to gain some for himself by singing the French song Alouette, but he is hardly noted by his aunts. All they grunt to him is “You still here?” and, to themselves:

(11)?* The older they gpt, the cuter they ain't

31This utterance is perfectly understandable—it means “As they [i.e., children in general] get older, they aren't cuter”—but it is far from perfectly grammatical. (And the sentence does not improve if we replace the colloquial contraction ain't by the more standard form aren' t.) The near-ungrammaticality clearly arises from the use of the negative verb form. If we use a positive form instead, the sentence becomes fully grammatical (The older they get, the cuter they are). There is also nothing grammatically wrong with the following sentence:

(12) The older they get, the less cue they are.

32Although (11) and (12) are close in meaning, they are not semantically equivalent. This becomes clear when we consider the related head-clause first pattern (see again section 2.2., in connection with the examples in (6)). Surprisingly, this pattern does allow negation:

(13) They {ain't/aren't} cuter the older they get.

33We can see now that the “less cute” interpretation is not logically entailed but only implicated, which means that it can be canceled. For example, the speaker can continue (13) with... but they aren't less cute either—they're always as cute as can be.

2.3. 2. McCawley's take on the matter

34The observation that the actual correlative comparative construction, unlike the related head-clause first version, seems to disallow negation of the head-clause has already been made by McCawley (1988a, 178-179). He maintains that these characteristics need not be stipulated, since they follow from general syntactic principles. McCawley's account, however, is couched in heavily transformational terms and therefore lacks intuitive appeal. It seems to me quite impossible to strip his explanation of all framework-specific terminology and rephrase it in more general, theory-neutral terms. One aspect that might nonetheless be transferred is the idea that the negator not is underlyingly outside (at the left of) the sentence but obligatorily moves to a position after the subject; this movement is then blocked in correlative comparatives (for reasons that are only valid within McCawley's transformational approach and therefore do not concern us here). What strikes me as correct, at any rate, is that it is grammatically as well as notionally possible to negate the proposition expressed by a correlative comparative construction if the negator not is left completely outside the construction:

(14) It is not the case that the older they get the cuter they are.

2. 3 .3. When negation is possible in the actual correlative comparative construction

35It is important to realize that negation of the head clause of the correlative comparative construction is not altogether disallowed. Here are some authentic examples in which the negator appears on the verb of the head clause without causing ungrammaticality:

(15) a The more I try to talk to him the more he doesn't listen...
b.... the harder [I] try the more [I] can't get to sleep...(
c. As an old wise saying has it, “The more you know, the more you don't know.”(

36What is curious, now, is that the following related head-clause first versions are not equivalent to their respective counterparts in (15):

(16) a. He doesn't listen more the more I try to talk to him. (≠ (15a))
b.... I can't get to sleep more the harder I try... (≠ (15b))
c. You don't know more the more you know. (≠ (15c))

37The reason for this semantic non-equality is this: in (15), the comparative phrase in the head clause has scope over the negator, while in the head-clause first versions in (16), it is the negator that has scope over the comparative phrase, which is fully integrated in the head clause. In other words, the sentences in (15) can be schematized as (17a), and those in (16) as (17b):

(17) a. “The more X, the more it is the case that not Y”
b. “It's not the case that Y more the more X”

38In order to provide semantically equivalent head-clause first versions for (15a-c), we need to find a way of bringing the negator within the scope of the comparative phrase, so that the sentences are of the conceptual structure “It is more the case that not Y the more X.” This can be achieved if we depart slightly from the standard form, but the result is never completely felicitous, probably because principles of information packaging (rather than of pure syntax) are being violated then. For example:

(18) a. ? He increasingly doesn't listen the more I try to talk to him.
b. ?... I more and more can't get to sleep the harder I try...
c. ? There are more and more things you don't know the more you know.

39In any case, it is clear that the actual correlative comparative construction (the more X, the more Y) fully allows negation of the head clause as long as the negation is conceptually included in the scope of the increase (or decrease). This condition is not fulfilled in (11), which paraphrases as:

(19) “As they get older, it's not the case that they are correspondingly cuter.”

40From this paraphrase, it is evident that the negation has scope over the increase instead of the other way round—hence the very marginal acceptability of the correlative comparative construction.

41Observe, further, that negation in the head clause is automatically within the scope of the comparative phrase if this comparative phrase (minus the) is not a complement of the verb. Apparently, a comparative phrase that is not analyzable as part of the complementation of the verb has wide scope, and clause negation does not stand a chance of out-scoping it. By contrast, if the comparative phrase is included in the verb phrase (as a complement of the verb), negation takes wide scope over it. For example, because listen is an intransitive verb—you cannot * listen something—the second occurrence of more in The more I talk to him, the more he doesn't listen cannot possibly be analyzed as a complement of listen and, accordingly, the negation is included in the scope of the increase. Compare with... the cuter they are, where cuter functions as a subject complement. Accordingly,...?* the cuter they aren't is hardly acceptable, since the negator includes the comparative phrase in its scope.

42A more intricate case is The more I know, the more I don't know. Here, more can, in principle, be analyzed as a complement of the verb know—this is, in fact, the only acceptable parsing for the subordinate clause. However, in the head clause, more is not a complement of know but a complement of the understood existential verb be:... the more (there is that) I don't know. (Similarly, the sentence The more I know, the less I don Ί know clearly does not mean “As I know more, I don't know correspondingly less,” but “As I know more, there is correspondingly less that I don't know.”) The negator only applies to know and not to the superordinate (though unexpressed) existential verb, and by virtue of being a complement of a syntactically higher-ordered verb than know, the comparative phrase has scope over the negator.

2. 3. 4. Conclusion

43The main clause of a correlative comparative construction does not readily allow negation. Only if the negation falls within the scope of the comparative phrase does the construction sound fully acceptable. Compare:

(20) a. The older they get, the cuter they are.
b. ?* The older they get, the cuter they aren' t.
(21) a. The older they get, the more I like them.
b. The older they get, the morel don't like them.

44We have implicitly rejected the first claim presented in (3) above, which says that negation is a simple operation. Evidently, negation here is not such a simple operation, since the speaker cannot just insert the negator as usual but has to take into account the relative scope of the negator vis-à-vis the comparative phrase. Furthermore, the second claim in (3) has been implicitly refuted as well. Indeed, we are faced here with a construction that can contain a positive clause that does not have a negative counterpart.

45Importantly, negation has been shown in this section to be a feature that is not construction-independent. That is, speakers cannot just add or remove a negator as they please, because constructions can specify that they disallow or require negation—or allow it under certain conditions, as in the construction we have discussed in this section. In the following two sections, we will briefly look at negation in two other constructions in the grammatical periphery.


First Church
of Springfield
(Sign outside church in The Simpsons, episode “Bart sells his soul”)

3.1. An introduction to the pattern at hand

  • 4 I am grateful to Barbara Dancygier for sending me Chapter 9 of the book on conditionals she co-aut (...)

46The English grammar contains a parallel, conjunctionless pattern of the form No X, no Y, as illustrated in (22a-b) below, where X and Y are filled by different nouns (which may or not be inflected for plural), and for which a conditional paraphrase can be given (e. g. “If there is no pain, there is no gain,” that is, “If you aren't prepared to suffer, you won't achieve anything”). This pattern is briefly discussed in Declerck and Reed (2001, 407), who refer to it as an “asyndetic paratactic conditional” or a “juxtaposition conditional,” and more fully by Dancygier and Sweetser (in press, 214-215), who discuss it in a section dealing with “non-clausal coordinate conditionals without conjunctions.”4

(22) a. No pain, no gain.
b. No guts, no glory.
c. No risk, no reward.
d. No money, no honey.
e. No work, no pay. (and also, from the perspective of disgruntled employees: No pay, no work.)
f. No shoes, no shirt, no service. (or: No shirt, no shoes, no service.)
g. No pain, no palm; no thorns, no throne; no gall, no glory; no cross, no crown (William Penn)

47The above examples are all lexicalized to a high extent. They are part of the conventional stock of fixed English expressions, including quotations. These conventionalized expressions probably make up the bulk of the occurrences of the No X, no Y pattern, but novel, productive instantiations are fully allowed as well (e. g. No car, no job or No oil, no economy).

48The last but one example (No shoes, no shirt, no service) is a common dress code sign at the entrance of restaurants in the States. Here, the first two conjuncts function as protases (antecedents) with respect to the third conjunct, which functions as the apodosis (the consequent). The bitingly humorous epigraph to this section is only one example of the popularity of varying on this standard formula. In this satirical mimicry, the Church of Springfield is painfully revealed as failing to obey God's command to help the needy (“clothe the naked”) and as denying Jesus' maxim that “the last [i. e., the poor and needy] will be the first [to be saved]”. Dancygier and Sweetser (in press, 215) also mention the following concessive variants, where the original version is playfully mocked:

(23) a. No shoes, no shirt, no problem!
b. No shoes, no shirt, full service!

3.2. The lack of a positive counterpart

49Although No X, no Y implicates that “if there is X, there is (or will be) Y”, the positive version does not seem to work. So, instead of No pain, no gain, we could not (or hardly) utter?* Pain, gain—unless we deliberately use our linguistic awareness to deviate from conventional usage—to mean “If you are prepared to suffer, you will achieve something.” This may be due to the fossilized status of the original version, but it is clear from? A car, a job or, worse,?* Oil, economy that the conditional interpretation is only conventionally conveyed when the nouns are preceded by no. Moreover, the normal interpretation of a conditional of the type ~ P ~ Q (i. e., If not-P, not-Q) is that Q is (or will be) true only if Ρ is true. This conditional strengthening cannot be occasioned by the corresponding positive version, if this version is at all acceptable. As Dancygier and Sweetser (in press) put it:

“Interestingly, the positive counterparts of some of the negative juxtapositions above seem quite incoherent: a sign such as (a) shirt, shoes, service would not successfully convey that customers will be served if wearing a shirt and shoes. At least part of the difference between the interpretation of negative and positive juxtaposed phrases seems to be regularly attributable to the negative semantics. No X, no Y, glossed as” If there is no X, there is/will be no Y, “expresses an only-if relationship: if ~ P, ~ Q means that Q can (at best) hold only if Ρ does. But a positive pairing of NPs does not seem conventionally interprétable by a similar process as meaning” Q only if P “. (Some) pain, (some) gain does not mean” there is some gain only if there is some pain. “If anything, one would interpret it as meaning something like” if (but not only if) there's some pain, there's some gain “. And even that is not a fully conventional reading like the negative NP pairings. This limitation is unique to this construction, since a positive if-conditional can easily be given an iff interpretation: for example,” If you are wearing a shirt and shoes, our staff will cheerfully serve you “can convey that you will be served if and only if you are properly dressed.” (Dancygier and Sweetser, in press, 214-215)

3.3. Some similar patterns, and what they lead us to conclude

50The No X, no Y conditional is not unique in lacking a positive counterpart. The (a)-sentences below illustrate three more constructional patterns for which a positive version is unavailable, and hence prove wrong the third claim in (3), according to which all negative clauses have a positive counterpart:

(24) the Just because X doesn't mean Y construction:
a. Just because it's herbal doesn't mean it's safe.
b. * Just because it's herbal means it's safe. (25) the If it (wasn't/weren't/hadn't been} for X, Y would Ζ construction:
a. If it hadn't been for her, I would have given up.
b. * If it had been for her, I would have given up (26) The epistemic can't construction:
a. He can't have killed her. (He has a solid alibi.)
b. * He can have killed her. (He doesn't have a solid alibi.)

51We see that for some constructions, negative polarity is an inherent, unalterable feature.


Homer: [sarcastically] Oh, I don't get the job, do I?
We-ell boo-hoo! I don't get to be a cartoon dog! (The Simpsons, episode “The Itchy and Scratchy and Poochy Show”)

4.1. An introduction to the constant polarity tag and its use in discourse

52The most typical question tag in English is the reversed polarity tag, which alters the polarity of its anchor clause:

(27) a You're tired, aren't you?
b. She wouldn't do that, would she?

53Less familiar is the so-called “constant polarity tag” (e.g., Huddleston 2002a, 895; Kimps 2003; see this latter study for further references and for alternative terms used in the literature, such as “reduplicative tag,” “copy tag,” “same polarity tag question”, etc.). This kind of tag is exemplified in the following dialogue (italics are added for clarity):

(28) Lady Callender: He wasn't long in India; he threw up his post twelve years ago, when his wife died
Guy: Oh, he's a widower, is he?
Lady Callender: Didn't you know?

Guy: Never thought about it, Aunt Peggotty—or I should have said he was a bachelor; he has the cut of one.

54There are three distinctive properties of this kind of question tag. First, the tag typically has a slightly rising tone, but it does not always constitute a separate intonation unit (hence the frequent omission of a preceding comma in writing). Second, the anchor clause of the tag contains information that is repeated or, as is the case here, inferred from what has just been said by the interlocutor. Third, the tag is used to indicate that the proposition of the anchor clause is somewhat unexpected to the speaker or, at any rate, to. express recognition of the news value of the information. The tag is, in other words, a marker of mirativity:

“The term” mirativity “refers to the linguistic marking of an utterance as conveying information which is new or unexpected to the speaker”
(DeLancey 2001,369-370).

55In (27), there is genuine (if mild) surprise at what is said. Often, the speaker's surprise at the interlocutor's message is feigned and/or mingled with other illocutionary meanings. As Huddleston (2002a, 895) writes, “One use, commonly accompanied by so or a comparable item such as oh, I see, etc., carries an emotive meaning of disapproval, reproach, belligerence, or the like.” And “[b] ecause the anchor proposition is implicitly attributed to [the interlocutor], this use lends itself to sarcasm, as when I say to someone who has performed badly: So you're the one who was going to come back laden with prizes, are you?” (ibid.). (See, among others, Kreuz and Glucksberg, 1989, for the echoic nature of sarcasm.) In her detailed study of constant polarity tags, Kimps (2003, 62) states that

“the concept mirativity is accompanied by a few subtypes or interpersonal attitudes, such as surprise, doubt, disbelief, disagreement, disapproval, etc. These are often accompanied by undertones, such as irony, sarcasm, contempt, condescension and other hostile, aggressive, confrontational or challenging attitudes.” (Kimps 2003, 62)

56Here are some authentic examples of such antagonistic stances:

(29) a. Oh you think it's funny, do you?” she asked dangerously.
b. So you are an expert, are you? Your modesty is overwhelming.
c. Oh, I see. I'm the naive one,am I?

4.2. More on the relation between mirativity and antagonism

57Cappelle (2003) discusses a dialectal dialogic particle in Dutch that is also tagged to an echoic or inferential anchor and that also can convey, apart from some surprise at the contents of the anchor clause, a number of antagonistic emotive stances, such as mockery, reproach or threat. This suggests that there may be a cross-linguistic pattern at work, such that mirativity markers easily adopt belligerent shades of attitudinal meaning. In any case, the similar illocutionary force of this dialectal particle lends further support to Kimps's account, in which the mirativity sense is put forward as a core meaning and in which it is argued that “the attitudinal senses or overtones emerge as contextualizations of the core meaning” (Kimps, 2003,62).

58Compare with Huddleston's (2002a: 895) account, in which the mirative use (“where I accept what you say”) and the antagonistic use (where I challenge what you say) are presented as two rather distinct uses of the constant polarity tag. Huddieston (ibid.), though, does acknowledge that both uses share an important feature, distinguishing them from reversed polarity tags: in either case, “the anchor proposition derives from the addressee, rather than representing a prior belief of the speaker” (cf. the second characteristic we mentioned above, in section 4.1).

4.3. The constant polarity tag after a negative anchor

59What happens, now, if the anchor clause is negative? Given the term “constant polarity tag,” we would expect the tag to retain the negative polarity of the anchor. However, as Huddleston (2002a: 895) points out, many speakers only accept these tags when they are positive. Likewise, McCawley (1988b) writes:

“Examples like [? So John isn't arriving until 7:00, is he? Well, that's too bad.], while slightly odd, are much more normal-sounding than corresponding examples with a negative tag (* So John isn't arriving until 7:00, isn't he?) and are the only halfway normal form that a tag on a negative host can take in contexts that call for a reduplicative tag. Note that in this case there is no difference in surface form between reduplicative and reversal tags.” (McCawley 1988b, 499, note 7)

60The epigraph to this section contains an authentic example of a positive tag attached to a negative anchor in a setting that clearly begs for a tag of the sort we are concerned with:

(30) [Homer Simpson is auditioning for the voice of a cartoon dog with attitude ]
Homer: [clears throat] Ruff, ruff! I'm Poochie, the rockin' dog!
Myers: Now, that's just bad. You've got no attitude, you're barely outrageous, and I don't know what you're in, but it's not my face. Next!
Homer: [angry] Oh, no attitude, eh? Not in your face, huh? Well, you can cram it with walnuts, ugly!
Myers: That's it! That's the Poochie attitude, do that again!
Homer: Huh? I can't, I don't remember what I did. Myers: Then you don't get the job. Next!
Homer: [sarcastically] Oh, I don't get the job, do I? We-ell boo-hoo! I don't get to be a cartoon dog! Myers: That's if you've got the job!
Homer: [still sarcastic] Oh, now I've got the job, huh? [quietly] Oh, thank you.

61The intonation is rising, as can be verified when we watch the episode in question, the anchor clause is echoic, and the tone is unmistakably hostile and sarcastic—something which the transcribers at, where the fragment has been taken from, have not failed to notice, witness their added stage direction for Homer's line.

62This means that “constant polarity tag” is something of a misnomer. To the extent that speakers prefer to use a positive tag even after a negative anchor, I offer the term “constant positive polarity tag” as a more adequate (though somewhat clumsy) alternative: the polarity of the tag under discussion here remains positive (and is therefore, in a sense, constant), irrespective of the polarity of its anchor.

4.4. An evolution in the English language?

63I hypothesize that there is, or has been, a grammatical shift in the English language from “constant polarity tags” to “constant positive polarity tags.” It is reasonable to assume that positive anchors have always outnumbered negative ones, so that the majority of the actual tags (of the relevant kind) have always been positive. This way, the positive form of the tag has had more opportunity to become entrenched and fossilized.

64Evidence for this hypothesis comes from web searches for sentences like Oh, I don't, don't I? versus Oh, I don't, do I? Because the tag under consideration always reacts to something the interlocutor has just said we can expect the anchor to be frequently elliptical. This provides us with an easy way to collect and compare tags. It appears that most examples of the type Oh I don't, don't I? are found in texts dating from before 1950. Full details of this evolution await further research, but the following two examples seem to be representative of, respectively, an older and a newer stage of the English language:

(31) a.“... you don't leave this room until you have signed a paper that I have here on my desk.”
“Oh, I don't, don't I? Who will stop me?”
b. “How unpleasant could removing a scar be?” Ginny said, brow furrowing.
“Fairly,” Harry said, still in that dry tone. “You don't want to know the details.”
A flash of irritation shone in her eyes. “Oh, I don't, do I?” she snapped. “So glad you could read my mind for me, Harry, so I know what I do and don't want to know. Makes things so much easier on me.”

65The first example comes from Sir Arthur Conan Doyle 1926 book The Land of Mist; the second example comes from a present-day Harry Potter fan-fiction site.

4.5. Some similar patterns, and conclusions

66There does not seem to be any semantic difference between the negative and the positive version of the tag under discussion after a negative anchor—the only difference is stylistic, in that the latter probably sounds rather archaic and is (therefore) not acceptable for all speakers of English. Disregarding this stylistic difference, this tag may be counted among “a handful of English constructions in which, quite surprisingly, one can add or remove a negation without change of meaning,” as reported by Potts (2004):

  • 5 See also Cappelle (2002, 379-380), where I discuss the following patterns: (i) a. Otto is a person (...)

(32) a. I (don't) know squat (or: jack, beans, diddley) about phrenology.
b. That'll teach you (not) to tease the alligators.
c. I wonder whether we can ('t) find some time to shoot pool this evening.
d. You shouldn't play with the alligators, I (don't) think.5 e. I could (n't) care less about monster trucks.
(Potts 2004)

67I do not think that these “negation-indifferent” patterns seriously undermine the general definition of negation attempted in section 1.2. above. Although it is true that for these cases, adding a negator fails to result in a semantically opposite proposition (if indeed a tag can be said to express a proposition at all), it is clear that the remarkable property of resisting semantic change under a polarity reversal is construction-specific. In any case, these sentences again show that negation is not such a simple operation, contra the first claim in (3), in that its use and interpretation may turn out special in certain constructions.

68Furthermore, although after a negative anchor, the tag we have been considering in this section may allow both a positive and, for some speakers, a negative version (e. g. Oh, I don't know what I'm talking about {do/% don't} I?), a positive version is the only choice we have after a positive anchor (e. g. So you're too good for me, {are/* aren't} you?). Consequently, if we consider the tag itself to be a clause, then the second claim (saying that each positive clause has a negative counterpart) does not hold true for this antagonistic tag. Nor indeed do the second claim and the third claim (together saying that each clause has both a positive and a negative version) hold true for the ordinary reversal tag (e. g. It's rather cold, {* is/isn't} it?] You won't tell anyone, {will/* won t} you?).


Bart: Dad, are you licking toads?
Homer: I'm not NOT licking toads. (The Simpsons, episode “Missionary: Impossible”)

69In an ideal grammatical world, inserting a negator nicely reverses the truth value of a proposition. Inserting another negator then logically has the effect of canceling this truth reversal—beside creating intentional confusion, as in Homer's reply above. This situation may be valid for standard clause types in English, but—fortunately, perhaps, for grammarians—reality is not as simple as that. In this paper I have shown that there are constructions in the periphery of grammar for which negation is not just a feature that one is free to add (or remove). In one construction, the correlative comparative, negating the main clause is hardly acceptable (e. g. ?* The older they get, the cuter they aren't), unless the negation is included in the scope of the comparative phrase (e. g. The older they get, the more I don't like them = “The older they get, the more it's the case that I don't like them”). In another construction, the No X, no Y pattern, the negation in both halves is already part of the package and cannot be removed (e. g. *(No) oil, *(no) economy). In yet another construction, the “constant (positive) polarity tag,” (as in, e. g., Oh, so you want to play it like that, do you?) the polarity is the same as the polarity of the clause to which the tag is anchored, but there are indications that the polarity of the tag increasingly tends to remain positive even after a negative anchor clause (e. g. I see, I'm not good enough for you, am I?).



BECK Sigrid, “On the semantics of comparative conditionals”, Linguistics and Philosophy, 20, 229-271, 1997.

BORSLEY Robert D., “An approach to English comparative correlatives”, Stefan Müller (ed.), Proceedings of the 11th International Conference on Head-Driven Phrase Structure Grammar. Stanford: CSLI Publications, 70-92, 2004a.

BORSLEY Robert D., “On the periphery: Comparative correlatives in Polish and English”, Proceedings of Formal Approaches to Slavic Linguistics, 12, 59-90, 2004b.

CAPPELLE Bert, “On the status of intervening think-clauses in English long-distance dependency constructions”, Leuvense Bijdragen (Leuven Contributions in Linguistics and Philology), 91 (3-4), 369-381, 2002.

CAPPELLE Bert, “Het gebruik van het Brugse dialoogpartikel (/dae/)”, Neerlandistiek. nl, 3 (1), 2003.(

CULICOVER Peter w. & Ray JACKENDOFF, “The view from the periphery: The English comparative correlative”, Linguistic Inquiry, 30, 543-571, 1999.

DANCYGIER Barbara & Eve SWEETSER, Mental Spaces in Grammar: Conditional Constructions (Cambridge Studies in Linguistics). Cambridge, Cambridge University Press, in press, 2005.

DECLERCK Renaat & Susan REED, Conditionals: A Comprehensive Empirical Analysis (Topics in English Linguistics). Berlin and New York, Mouton de Gruyter, 2001.

DELANCEY Scott, “The mirative and evidentiality”, Journal of Pragmatics, 33, 369-382, 2001.

DEN DUCKEN Marcel, “Comparative correlatives comparatively”, Unpublished paper, CUNY, 2003.(

FILLMORE Charles J., “Varieties of conditional sentences.” Eastern States Conference on Linguistics, 6, 163-182, 1986.

FRIED Mirjam & Jan-Ola ÖSTMAN, “Construction Grammar: A thumbnail sketch,” Mirjam Fried and Jan-Ola Östman (eds.), Construction Grammar in a Cross-Linguistic Perspective (Constructional Approaches to Language). Amsterdam and Philadelphia, John Benjamins, 11-86, 2004.

HUDDLESTON Rodney, “Clause type and illocutionary force”, Huddleston and Pullum (eds.), 851-948, 2002a.

HUDDLESTON Rodney, “Comparative constructions”, Huddleston and Pullum (eds.), 1097-1170, 2002b.

HUDDLESTON Rodney & Geoffrey K. PULLUM (eds.), The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press, 2002.

KIMPS Ditte. “Constant polarity tag questions: State of the art and descriptive analysis of its form, meaning and discourse uses.” Unpublished Master's thesis. K. U. Leuven, 2003.

KREUZ Roger & Sam GLUCKSBERG, “How to be sarcastic: The echoic reminder theory of irony”, Journal of Experimental Psychology: General, 118, 374-86, 1989.

LAKOFF George, Don't Think of an Elephant! Know your Values and Frame the Debate. White River Junction: Chelsea Green Publishing, 2004.

MCCAWLEY James D., “The comparative conditional construction in English, German, and Chinese.” Proceedings of the Annual Meeting of the Berkeley Linguistics Society, 14, 176-187, 1988a.

MCCAWLEY James D., The Syntactic Phenomena of English. Volume 2. Chicago and London: The University of Chicago Press, 1988b.

POTTS Christopher, “Negated, or not.” Language Log, 21 January 2004.(itre. cis. upenn. edu/~ myl/languagelog/archives/00 0368. html).

MICHAELIS Laura Α., “A case of constructional polysemy in Latin.” Studies in Language, 18, 1,45-70, 1994.

SAVELLI Marie-Josée, Contribution à l'analyse macro-syntaxique: les constructions ‘siamoises’ du type: plus v1, plus v2. Ph. D. dissertation. Université de Provence, 1993.

SWEETSER Eve, “Negative spaces: Levels of negation and kinds of spaces.” Talk given at the international conference on “Negation: form, figure of speech, conceptualization.” Tours, 8-9 October, 2004.

THIERSCH Craig, “The harder they come...: a note on the double comparative construction in English”, Werner Welte (ed.), Festschrift für Alfred Wollmann. München: Hueber, 47-65,1982.


1 For a full and annotated list of blackboard quotes, see The Simpsons Archive on

2 Negation should be seen here as shorthand for ordinary clause negation. I am not concerned with negation that applies to subclausal elements (e. g. He played a not insignificant part in that decision).

3 As is customary in linguistics, curly brackets (“{}”) enclose linguistic alternatives, separated by a forward slash. An asterisk (“*”) marks an entity (be it a word, a word group or an entire sentence) as downright ungrammatical. A question mark is used to indicate that the word (group) or sentence thus marked is of rather questionable grammaticality, without being wholly unacceptable. The combination “?*” is used to indicate an entity as verging on the ungrammatical. We also use the sign '%' to indicate that an entity is only acceptable for some groups of speakers (sharing the same language variety). Note, further, that any of these signs placed outside brackets indicates the grammaticality status of a sentence in which the bracketed material is left out.

4 I am grateful to Barbara Dancygier for sending me Chapter 9 of the book on conditionals she co-authored with Eve Sweetser. The reader should be aware that references to and quotations from this chapter should be treated with some caution, as the final version may differ from the pre-final version I used.

5 See also Cappelle (2002, 379-380), where I discuss the following patterns: (i) a. Otto is a person who I think's ready for such a challenge.
b. Otto is a person who I don't think's ready for such a challenge.
c. Otto is a person who's not ready for such a challenge, I don't think.
In the first sentence, the relative clause has positive polarity; in the second sentence, with an added negator, it has negative polarity; in the third sentence, it still has negative polarity, despite the fact that there is one more negator. The tagged negative think-clause here does not cancel the negative polarity of the relative clause, but, as it were, merely copies its polarity.


Is a post-doctoral researcher at K.U. Leuven, where he obtained a Ph.D. in linguistics with a thesis on verb-particle patterns in English. At campus Kortrijk of K.U. Leuven he currently works together with Renaat Declerck and Susan Reed on a project resulting in a multi-volume grammar of the English verb phrase. While his main interest remains with particle constructions, he sporadically writes about other constructional phenomena

© Presses universitaires François-Rabelais, 2006

Conditions d’utilisation :