Towards a taxonomy of linguistic jokes. (1).

1. Introduction

The range of criteria that have been used for classifying verbal jokes is large and includes, among others, such terms as theme, subject, cycle, target (butt), level of propriety, origin, narrative form, and length of the text. All types of classification have their own validity as well as their own practical uses (for instance, for public speakers wishing to embellish their speeches with a joke or two a classification by subject or theme might be particularly useful, etc.)

One persistent problem in the classification of jokes has been the urge to apply multiple mutually incompatible criteria to produce a single classification (see Raskin 1985: 29 for a critical assessment of such taxonomies). Even if a single classification criterion can be agreed on, a further problem arises of the choice of values for the classification variables: typically, they come from open, unconstrained sets. Thus, there may be little correspondence between two classificatory efforts even if they claim to use the same criterion. For instance, Stanley (1980) has subject categories such as "teaching terrors" or "in the drink", which are absent from Johnson (1989), while the latter has "sex education" and "politics", none of which appears in the former collection. This particular problem seems to be related to the predominantly semantic nature of the classificatory criteria, and it reflects the open-endedness of semantic systems. On a more general level, the enormity of the problem of classifying jokes in a principled and uniform manner is caused by the richness of types of humour tokens, including verbal jokes. It may thus be impossible in principle to usefully categorize jokes along a single parameter. A recent multi-parameter joke model proposed by Attardo - Raskin (1991) holds some taxonomic potential, but fails to capture the distinction, important for this author as well as for many others (Milner 1972, Hockett 1973, Shultz -- Horibe 1974, Frumusani 1986, Shultz - Robillard 1980, Skowronska 1989, Spector 1990, Attardo 1993, Attardo 1994, Lew 1996a, 1996b), between linguistic and non-linguistic jokes. For the basis of this distinction and arguments in its favour see Lew (1996a, 1996b).

Given the complexity of the phenomenon of humour and the daunting variety of joke types, I think it may be beneficial to build taxonomies of specific separate types of jokes. Within the smaller types, uniformity of classification criteria should be easier to achieve. One such type is linguistic jokes. As shown in Lew (1996a, in press), linguistic jokes may depend on linguistic ambiguity, understood as

that property of a fragment of text which allows for two or more significantly different semantic interpretations to be arrived at by a substantial proportion of typical text recipients. (2)

Attardo et al. (1994) have found that 431 of the 441 verbal (this term roughly corresponds to my linguistic) jokes, that is 98%, were ambiguity-based, within a total sample of 2000 printed jokes (linguistic or otherwise). Although Attardo et al. (1994) did not probably use "ambiguity" in an equally broad sense as here, this difference should not affect the qualitative implications of the finding, namely that very nearly all linguistic jokes are ambiguity-based. Such jokes may be classified according to the type of ambiguity that they involve, and this is the principal aim of the present paper. Depending on the length, extent, or, more precisely, status within the linguistic system of the fragment of a joke's text that is open to two radically different interpretations, more than one type of ambiguity can be distinguished, in consequence yielding different classes of ambiguity-based linguistic jokes.

Apart from a purely theoretical-descriptive interest, there are important practical benefits of having such a classification which go beyond humour studies per se. For example, jokes classified by the type of ambiguity involved can be, and have been, used to test the perception of ambiguity in children, thus contributing to research on language acquisition and development (Shultz - Pilon 1973, Shultz - Horibe 1974, Fowles - Glanz 1977, Shultz - Robillard 1980, Horgan 1981, Hirsh-Pasek - Gleitman - Gleitman 1986, Sinclair - Jarvella - Levelt 1986, Klein 1992). They can also be used to test the perception of ambiguity in language-impaired individuals, thus assisting in the research on language deficits (Spector 1990). In such studies, whether and when the ambiguity is perceived can be tested by checking if, and under what conditions, subjects get the jokes. In order for such studies to be valid and maximally profitable, however, the joke tokens used in the studies need to be classified consistently and correctl y. As I have demonstrated elsewhere (Lew 1996a: 97103), the reverse has usually been the case.

2. Types of jokes

2.1. Lexical jokes

2.1.1. Polysemous lexical items, homonyms and homophones

It is no coincidence that the discussion begins with this category. In English, lexical jokes appear to be by far the most frequent single category of linguistic jokes. In the aforementioned study by Attardo et al. (1994), 426 of the 441 jokes classified there as verbal, that is 96%, were lexical.

In a lexical joke, the ambiguous string is coextensive with a lexical item in the text of the joke. While much has been made in the literature of the differences between polysemy, homonymy, and homophony, all these cases share the above-stated essential feature. Since the distinction between polysemy and homonymy is often problematic, and homophony is for some authors just a special case of homonymy (McArthur 1992), to me it does not seem a fruitful venture to try and tease the three apart here. However, it would not be difficult to further subdivide the present category along these lines, should one wish to do so.

Consider joke (1) below:

(1) "Have you ever appeared as a witness in a suit before?" asked the judge.

"Why of course!" replied the young girl.

"Will you please tell the jury what suit it was?"

"It was a pink suit," she replied quickly, "with red collar and cuffs, and buttons all the way down the front."

(Misztal 1990: 786)

Joke (1) is a typical example of a lexical joke. The two characters of the joke, the judge and the young girl appearing on the witness stand, apparently interpret the word-long string suit in the two questions of the judge as 'lawsuit' and 'type of attire', respectively. Initially, it is the judge's interpretation (original interpretation) that is salient and exclusively accessible to the typical recipient (i.e. hearer or reader, as the case may be) of joke (1). The second interpretation (alternative interpretation) becomes accessible when the girl starts elaborating on the physical description of the suit she was wearing on the other occasion. Once the alternative interpretation becomes accessible, it creates ambiguity -- and the humour -- of the joke. While the fact that it is the word suit that is the source of ambiguity may appear obvious, it is possible to put this fact to test by altering parts of the text of the joke. We should be looking for the minimal alteration that would remove the ambiguity. Subs tituting a near synonym for the suspected candidate that does not allow the same kind of polysemy or homonymy will normally point to lexical ambiguity created by that candidate. Replacing all occurrences of suit with trial, as in (2) below, will remove the ambiguity, and make the girl's response incongruous:

(2) "Have you ever appeared as a witness in a trial before?" asked the judge.

"Why of course!" replied the young girl.

"Will you please tell the jury what trial it was?"

"It was a pink trial," she replied quickly, "with red collar and cuffs, and buttons all the way down the front."

The reader has to take into account that having just been exposed to the original joke (1), the word suit with its two interpretations will still be salient in the reader's memory. Without such priming, however, there is no ambiguity in (2), unless we settle for the Humpty-Dumpty type of "private meaning", in which case for the girl trial could refer to something she wears. Even when saying this, however, we are making a statement about a word-long string! It is important to realize that the proposed synonym-substitution procedure can locate the ambiguous string with a large degree of objectivity and algorithmicity. Although here I went straight for the actual ambiguous lexical item, a rigorous procedure would consist in a left-to-right (or vice versa) item identification and substitution. I will spare the reader this, but the reader is invited to check some of those substitutions at least (eg. inquired for asked, justice for judge, court for jury, blue for pink, responded for replied etc.) and verify that th ey, unlike the substitution of trial for suit, will not rid the original joke of its ambiguity or humour.

Replacing suit with dress (roughly the girl's interpretation), as in (3) below, will again remove the ambiguity:

(3) "Have you ever appeared as a witness in a dress before?" asked the judge.

"Why of course!" replied the young girl.

"Will you please tell the jury what dress it was?"

"It was a pink dress," she replied quickly, "with red collar and cuffs, and buttons all the way down the front".

In this case, there can be only one interpretation of the word dress, and so no ambiguity is present.

2.2. Lexicalization of a larger unit (lexico-syntactic)

In some cases the ambiguity of a joke may stem from a string being interpretable as either a regular phrasal unit, or as a more or less lexicalized item of a more immutable, close-knit type. Joke (4) is one case in point:

(4) A religious and charitable woman noticed a very down-and-out sort of man standing at the corner of the street near her residence.

One morning she took compassion to him, pressed a dollar into his hand and whispered: "Never despair."

Next time she saw him he stopped her and handed her nine dollars.

"What does This mean?" she asked.

"It means, ma'am," said the man, "that Never Despair won at 8 to 1."

(Misztal 1990: 611)

The woman means never despair as an expression of reassurance to who she takes to be a homeless beggar. Syntactically, on this reading, never despair is a clause. The man, who turned out to be an unlicensed bookie, interpreted never despair as a lexical unit being a proper name referring to a specific racing horse.

2.2.1. Decomposition of idioms

Whether, and to what extent, idioms should be treated as autonomous lexical items is debatable (for an overview and assessment of psycholinguistic models of idioms see Cieslicka-Ratajczak in press; see also Makkai 1972). Furthermore, not all idioms exhibit the same degree of linguistic autonomy. Assuming that idioms may, to some extent, be treated as separate lexical items, it seems appropriate to treat the present class of jokes as a subcategory of the lexico-syntactic class. Idioms provide a rich source of linguistic ambiguity in jokes. The meaning of an idiom is not easily predictable from the individual meanings of its elements. By interpreting a string coextensive with an idiom literally as being a simple product of the meanings of its elements an alternative reading is produced, different from the conventionalized idiomatic reading, and this generates ambiguity. Consider joke (5) below:

(5) Recruit, after physical: "Well, Doc, how do I stand?"

Doc: "I don't know. It's a mystery to me!"

(Hoke 1965: 118)

This simple joke well illustrates the operation of idiom decomposition. The recruit enquires about his state of health or possibly suitability for military service with an idiomatic clause how do I stand? The doctor interprets the clause literally (and possibly jokingly) as an invitation to explain the mechanism of the recruit's being able to maintain a vertical position with an implication that the recruit's physical condition is so serious (as a result of an illness, or intoxication), that one would not expect him to be able to stand on his feet.

Another quite similar and equally simple joke is (6):

(6) "I have changed my mind."

"Thank Heavens! Does it work any better now?" (Misztal 1990: 810)

The first character of (6) announces to the other character that (s)he has changed his/her mind, and the recipient of the joke interprets this to mean (presumably in accordance with the intent ascribed to the first character) 'reversed an earlier decision', which is a very common idiomatic expression. The second character's response compels the recipient to reinterpret changed my mind as a literal phrase meaning approximately 'had my brain replaced'.

That the ambiguity of such jokes indeed depends on the idiomatic interpretation of a phrase vis-r-vis a literal one can be verified by modifying the phrases slightly, so that the literal meaning remains roughly unaffected, but the idiomatic meaning, being (by definition) dependent on the precise wording, would be lost (i.e. it would not be likely to be accessed by the recipient of the joke). Thus, joke (5) above could be recast as (7) below:

(7) Recruit, after physical: "Well, Doc, how is it possible that I can still stand?"

Doc: "I don't know. It's a mystery to me!"

In (7), no idiomatic reading is possible, just the literal reading, and there is no ambiguity.

Similarly, joke (6) could be recast as (8):

(8) "I got a new brain."

"Thank Heavens! Does it work any better now?"

Again, only the literal reading is likely and there is no second, idiomatic reading. Joke (9) below is a somewhat less clear-cut case:

(9) Teacher: "So, who knows who Anne Boleyn was?"

Johnny: "I do, Sir! She was a pressing iron!"

Teacher: "She was a what?"

Johnny: "She was an iron."

Teacher: "Where did you get this idea from, Johnny?"

Johnny: "Well, Sir, it says right here in our history book: 'After divorcing

Catherine of Aragon, Henry VIII pressed his suit with Ann Boleyn."'

Johnny is apparently unfamiliar with the phrase to press one 's suit with somebody, meaning roughly 'to propose marriage to somebody'. Instead, he interprets pressed his suit with as meaning 'ironed his trousers and jacket with' and thus concludes that Anne Boleyn was an iron for pressing clothes. Whether the phrase is an idiom is an open question. It behaves grammatically more like an ordinary phrase than most undisputed idioms, being capable of a range of syntactic modifications (though not passivization: *his suit was pressed). Semantically, the meaning of the phrase rather transparently relates to some senses of press and suit which are used outside this phrase. On the other hand, the phrase's collocational restrictions are quite stringent and idiom-like. In place of pressed, one could say pleaded (or ple(a)d), but that just about exhausts the possibilities. Suit, it seems, is irreplaceable. Consequently, in (9) it may be possible to recognize ambiguity at the single word level as well as at the phrasal ( whole idiom) level. At the word level, the word suit is ambiguous between 'courtship' and 'trousers and jacket'; and the word press is ambiguous between 'plead' and 'iron'. On this interpretation, we would have two ambiguous lexical items. At the phrasal level the ambiguity is between the figurative and literal meaning of the idiom to press one's suit with somebody.

2.3. Syntactic jokes

Jokes based on syntactic ambiguity involve two readings that stem from two different syntactic representations. Following Huddleston (1984, 1988), I shall be making a high-order distinction between syntactic class and syntactic function. In consequence, syntactic jokes in my classification will be divided into two basic categories according to the way in which their syntactic representations differ: syntactic class jokes, in which the two readings differ in terms of their syntactic class, though they may (and often do) also differ in function; and syntactic function jokes, in which the two readings differ in terms of their syntactic function only.

2.3.1. Syntactic class jokes

Jokes based on syntactic class ambiguity exhibit two readings corresponding to two different syntactic representations, within which a fragment of text may be assigned two different syntactic class structures, and this fact underlies the difference between the two readings. Quite often, the placement of syntactic boundaries will be different in the two readings. Joke (10) below will illustrate the point in a simple way.

(10) Miss Wornout wrote on Bobby's report:

"Bobby's trying -- very." (Hoke 1965: 26)

The ambiguous fragment in joke (10) is trying. Initially, it is interpreted as a verb referring to Bobby's efforts in class -- a type of evaluation that one would reasonably expect a teacher to include in a pupil's report. Once very is processed, however, then to accommodate this afterthought-type additional material into the preceding material a reinterpretation is called for. Very cannot modify a verb phrase, nor can it modify a clause. Instead, it commonly modifies adjectives, and for this reason trying, on the alternative reading, is interpreted as an adjective in terms of its syntactic class.

Joke (11) further exemplifies syntactic class ambiguity:

(11) A homeowner in Minehead, England, wished to return an empty coal sack to his coal delivery man, so he left a note on the front door saying. "Empty sack in kitchen." When he returned he found a pile of coal on his kitchen floor.

On the first reading, and apparently in accordance with the homeowner's intentions, empty sack in kitchen would be interpreted as a noun phrase, or as an elliptical "telegraphic" declarative clause with the verb ('is') elided. On this reading, empty sack would be a noun phrase, with empty being an adjective (and an adjective phrase at the same time). On the second reading, which is triggered by the otherwise inexplicable action taken by the coal delivery man, empty sack in kitchen is an imperative clause headed by the verb empty.

A question arises at this point why we do not want to say that joke (11) is a lexical ambiguity joke, with empty being the ambiguous lexical item. After all, a selection of two distinct lexical items does seem to be involved. To make the distinction between the two types of jokes (lexical and syntactic) work, we need to postulate that the two readings in a lexical ambiguity joke have the same syntactic representation. This is consistent with the examples of lexical jokes I have provided above. Thus, in lexical jokes the two lexical items would exhibit differences primarily in their semantic, rather than syntactic, specification. Once the syntax is different across the two readings, it makes sense to give primacy to this difference, as a lower-level one. In addition, there is substantial overlap in the semantics of [empty.sub.Adj] and [empty.sub.V], since the latter is derived from the former in a synchronically rather transparent way: [empty.sub.V] = to make [empty.sub.Adj] -- still the more reason to classif y the joke as syntactic. Even in the absence of such close semantic correspondence, however, we will classify such jokes as (12) below as syntactic on the strength of the clear difference in the syntactic structures underlying the two readings:

(12) Buy your girl a bikini and watch her beam with delight.

The two readings of (12) are well paralleled by two versions with girl changed to boy plus other necessary modifications in (13) and (14) below, none of which exhibits ambiguity (with the added boon of balancing off some of the sexist content of (12) which certain readers may find offensive):

(13) Buy your guy a pair of tanga swimming trunks and watch him beam with delight.

as opposed to:

(14) Buy your guy a pair of tanga swimming trunks and watch his beam with delight.

The difference between the two readings is much more than a simple matter of lexical choice. The two altered versions of the original (12) involve a difference between him and his, which is clearly a grammatical difference. On the first reading of (12), parallel to (13), beam with delight is an embedded clause, with her possibly interpretable as subject On the second reading of (12), parallel to (14), there is no embedded clause, and her beam is simply a noun phrase functioning as direct object of watch.

Joke (15) below is a slightly problematic case:

(15) Fair Maid: "Oh, sir, what kind of an officer are you?"

Officer: "I'm a naval surgeon."

Fair Maid: "My goodness! How you doctors do specialize." (Pocheptsov 1974: 263)

On the face of it, the ambiguity of (15) seems to be a simple one between an adjective (naval) and a noun (navel), the two being phonetically identical. Joke (15) might thus be taken as a case of syntactic class ambiguity. However, it will be observed that the difference in syntactic class is not mirrored in a similar difference in syntactic function. On both readings, naval/navel functions as pre-modifier of the head noun surgeon within a noun phrase. In this function, nouns are deprived of some of their typical nominal properties. They don't take the usual range of dependents (I am a *a navel surgeon). They don't have a distinction in number. On the other side, naval is not gradable (a *very naval surgeon, a *more naval/*navaller surgeon). Thus, naval is not a prototypical adjective. The traditional division into syntactic classes (or 'parts of speech') is anything but watertight and does not produce homogenous categories (Huddleston 1984, 1988). In (15), the two readings are not really so distant syntactic ally. On the other hand, there is no doubt that we are dealing here with two completely unrelated lexical items. They are very distant semantically, unlike in (11). It might be argued that (15) bears a strong resemblance to (16) below:

(16) "Pardon me," said the stranger, "are you a resident here?"

"Yes," was the answer. "I've been here goin' on fifty years. What kin I do for you?"

"I am looking for a criminal lawyer," said the stranger. "Have you any here"?

"Well," said the other, "we're pretty sure we have, but we can't prove it." (Pocheptsov 1974: 264)

The ambiguity here is between criminal meaning roughly 'specializing in criminal law' and 'engaging in crime'. It should be clear from the foregoing discussion that (16) is a lexical ambiguity joke. On closer inspection of the syntactic properties of criminal within the two readings we find that in one of them it is gradable, but not in the other (and so a very criminal lawyer fixes the meaning and is not ambiguous). Furthermore, it is correct to say this lawyer is criminal in the 'engaging in crime' sense, but it isn't correct in the other sense. Thus, there are at least two syntactic differences between criminal on one reading and criminal on the other reading: one, but not the other, can be accompanied by adverbs of grade; and one, but not the other, can be used in the predicative function. There is a non-trivial syntactic difference between criminal of the two readings, and yet we classify the joke as lexical on the grounds of an important lexico-semantic distinction between the two readings. Likewise, (1 5) exhibits some (not of the first magnitude) degree of syntactic difference, and a very clear lexical one. Therefore, I believe that both (16) and (15) are most appropriately assigned to the lexical ambiguity class.

2.3.2. Syntactic function jokes

In jokes based on syntactic function ambiguity, the minimal ambiguous fragment of the joke's text is a syntactic constituent which exhibits two distinct readings differing in the syntactic function -- rather than syntactic class -- of the constituent (see also Bader 1994). Joke (17) is a case in point:

(17) Harriet: 'When I was a nine-year-old girl, I was left an orphan." Carrie: "Oh, yes. What did you do with it?" (Hoke 1965: 105)

While an orphan is a noun phrase on both interpretations, and thus the syntactic class remains unaffected, the syntactic function of this noun phrase is different within each of the two readings. On the original reading, apparently the one intended by Harriet, an orphan would be a subject predicative of a passive complex transitive construction (corresponding to object predicative of a nonpassivized clause -- see Huddleston (1984, 1988)). On the alternative reading, which is introduced by Carrie's response, the same NP, an orphan, would be interpretable as having the syntactic function of direct object. (3) The point in Carrie's response which triggers the alternative reading is the PP with it, with the introduction of the pronoun it. Assuming the complex transitive construction reading, this pronoun would have no likely candidate for antecedent within the whole context of the joke, since an orphan would be coindexical with I, and would refer to Harriet (from the particular point in time at which she became a n orphan). Since in What did you do with it? the you quite clearly refers to Harriet, it is inconceivable that it could refer to Harriet as well. The only plausible noun phrase with which it might be coindexed is an orphan, and this analysis leads into the alternative interpretation of the latter as a direct object, referring to a different entity than 1 of 1 was left an orphan.

A second example of a joke based on syntactic function ambiguity is given in (18) below:

(18) A man eating a kebab goes up to a lady who has a yapping chihuahua at her heels.

"Can I throw your dog a bit?" he asked politely.

"Certainly." came the reply.

So he scooped the dog up and flung it a hundred yards.

On the original reading, noun phrase your dog functions syntactically as indirect object, and noun phrase a bit functions as direct object, within a ditransitive clause. On the alternative reading, it's your dog that functions as direct object, while a bit functions as postmodifier of a verb phrase within a monotransitive clause.

2.4. Phonological jokes

The term "phonological joke" refers within the present taxonomic framework to any joke in which the ambiguous fragment of a joke's text might typically have non-identical phonetic forms for the two interpretations, though this typical non-identicality does not prevent the accessing of two readings by the joke recipient. Thus, getting a phonological joke may involve the recipient allowing two typically different phonetic strings to be jointly served by one phonetic form, which may be identical with one of the two, but it may also bridge the gap between the two by combining some elements of the two. This latter case usually requires a degree of skill in telling a joke, as proper attention must be given to the low-level phonetic detail of the joke's delivery in order to facilitate the accessing of both readings. One may, therefore, speak of the two different interpretations of the ambiguous fragment of the joke's text being phonologically mediated. The two readings of a phonological joke correspond to the source and target of a paronomasic (heterophonic) pun in the sense of Sobkowiak (1991). Quite beside phonetic non-identity, the two readings may yield different lexical items, or different syntactic representations, so it still makes perfect sense to speak of lexical, or syntactic, ambiguity. In essence, then, the division into phonological and non-phonological jokes is largely independent from the classification based on the type of ambiguity -- an important property that all previous attempts at classifying jokes that I am aware of have missed. A classification into phonological and non-phonological jokes may thus be said to cut across the primary type-of-ambiguity classification.

Joke (19) below will, I hope, well illustrate the point made above:

(19) Two boys were talking about their afterschool jobs.

Harry asked Larry: "How do you like your chimney-sweeping job?"

Larry answered smugly: "It soots me." (Hoke 1965: 108)

The two readings may be pronounced differently in some dialects of English, but will exhibit no phonetic difference in other dialects. Yet it is certainly not the case that (19) works well only in those dialects in which suits is pronounced [sju:ts] and soots is [su:ts]. Neither is it true that (19) works well only in those dialects in which the two are pronounced identically. In fact, it seems rather obvious that the joke works quite well, and pretty much in the same way, in both these classes of English dialects. It follows then that the (possible) non-identicality of sound is not the most essential feature of a joke as regards its potential to amuse. Far more important is the lexical ambiguity, that is the possibility of interpreting the fragment represented orthographically here as soots in two alternative ways: as 'soils with soot' and 'fits'. Thus, joke (19) above is primarily a lexical ambiguity joke, and, secondarily (in some dialects), a phonological joke.

Another example of a phonological joke is given in (20) below:

(20) Man: "I'd like to buy a pair of nylon stockings for my wife."

Clerk: "Sheer?"

Man: "No, she's at home." (Hoke 1965: 145)

On the original reading, apparently consistent with the clerk's intended meaning, the clerk's query is interpretable as an adjective, which would refer back to the stockings that the customer requested. On the alternative interpretation, apparently consistent with the customer's interpretation, what the clerk said would be something like (is) she here?, which would be an interrogative clause. Again, as in joke (19), the phonetic form of the two possible readings might in many circumstances be different, although it is conceivable that, in casual style, the two might be very similar or identical. Whatever the case, the essential thing for the recipient is accessing both readings, that is perceiving the ambiguity between the adjective (or adjective phrase) and the clause. (20) is primarily a syntactic class ambiguity joke.

Consider joke (21):

(21) A lady in a taxi arrives at a drugstore. She buys some cold tablets, pays and leaves -- leaving the medicine behind. She notices this when already in a taxi.

"Stop!" she shouts. "My aspirins!"

"No need to stop." replies the driver. "Just stick it out the window."

(21) is one example where skilful telling is crucial to the success of a joke. The joke should obviously be told in an American English accent. More specifically, the /p/ in aspirins should have an early onset of voicing, the word should be pronounced as just two syllables, and the vowel of the second syllable should be a centralized r-coloured vowel, rather than a sequence of [r] and a pure front vowel; thus, roughly [[LANGUAGE NOT REPRODUCIBLE IN ASCII]]. Unless sufficient care is paid to the low-level phonetic detail in telling the joke, the joke may not register with some audiences, and, admittedly, it is somewhat difficult to get joke (21) in its printed form.

I see no immediately obvious reason why recognizing phonetic non-identity should be limited to the segmental make-up of a given fragment of the joke, though in all studies dealing with "phonological" jokes that I am aware of it has been so limited. Suprasegmental properties, such as pitch variation, stress placement, and relative timing (rhythm) are all no less phonetically real (though perhaps less exhaustively researched and described in the phonetic literature), and reliable differences with regard to these properties can be ascribed to the distinction between the two readings of the ambiguous fragment of the joke.

Joke (22) below is a case in point:

(22) Haughty lady: "Must I stick the stamp on myself?"

Haughty clerk: "I think you'll accomplish more, madam, if you stick it on the package." (Hoke 1965: 148)

The ambiguity of (22) is in the clause must I stick the stamp on myself On the original reading, apparently consistent with the intended meaning of the lady character, on is part of the verb phrase stick the stamp on, whereas myself postmodifies this verb phrase. An alternative reading is revealed by the construction you stick it on the package, parallel in structure to that corresponding to the alternative reading. On this second reading, on myself is one complete constituent, a PP headed by the preposition on (in parallel to on the package) which functions as a complement within the verb phrase stick t on myself Within our framework, (22) will be classified as a syntactic-class ambiguity joke, as the ambiguous fragment has two distinct syntactic-class structures. However, the two readings are also typically associated with different phonetic realizations, which also qualifies (22) as a phonological joke. Whatever phonetic difference there is, however, cannot be expressed in terms of segmental composition, a s this is identical for the two readings (or, more accurately, whatever segmental variation is possible will not discriminate between the two readings). Rather, the phonetic difference is an intricate combination of timing and pitch variation, reflecting the syntactic difference between the two readings in the ambiguous fragment of the joke. Without going into too much detail as to what exactly the difference would be - that could only reliably be verified through instrumental study - let me briefly speculate on, roughly, what kind of differences one would expect to find. On the original reading, one would expect a prolongation of the syllable on, and a resetting (lowering) of pitch immediately following that syllable. On the alternative reading, one would expect a slowing-down on stamp, and a resetting of pitch to a lower level beginning with on, which would have a relatively shorter duration than the same word produced to represent the original reading, assuming that the overall tempo of speech would in bot h cases be similar. As in (21) above, so too in (22) skilful telling is of the utmost importance to the success of the joke. The teller should try to steer a middle course between the two phonetic forms so as to camouflage the typical difference between the two and thus improve the chances of the audience recognizing the ambiguity.

Perhaps an even clearer phonetic distinction which is not segmental in nature can be found in joke (23) below:

(23) "Waiter!"

"Yes, sir."

"What's this?"

"It's bean soup, sir."

"No matter what it's been. What is it now?" (Misztal 1990: 651)

In the printed version of the joke, the ambiguity is apparent from the spelling: bean versus been. It is not, however, just the choice between two different lexical items that forms the basis of the joke's ambiguity. The syntactic-class structure of (23) is radically different in the two interpretations, and on the strength of this we classify (23) as a syntactic-class joke (a further possibility that will actually be preferred for (23) is to recognize multiple dependent am biguity here: see section 2.10 below for a discussion of multiple ambiguity). On top of this, though, the two readings of the ambiguous fragment, it's bean soup, would typically yield different phonetic realizations. Quite apart from the segmental composition which may very well be identical (except that for some speakers, notably American, there might be a difference in the vowel between been and bean), the clearest difference would be in the placement of the nuclear, or tonic, stress, with all the consequences stemming from this (see Cru ttenden 1986). On the original "bean" interpretation, bean would receive nuclear stress. On the alternative "been" interpretation, soup would be the most likely item to take nuclear stress. Again, like (21) and (22) above, (23) calls for phonetically skilful telling. If a normal pattern compatible with the bean reading is used, it will produce an unreasonably high level of prominence on been within the "been" reading, and could prevent this particular reading from being recognized (effectively removing the humour), or would at least reduce the contextual justification for the second customer introducing the been reading, making the customer's response more strained within the overall structure of the joke, which detracts from the humorous value of the joke (Lew 1996a: 132-145). Probably the best line that a successful teller might adopt would be to give relatively equal prominence to both bean and soup, so as not to tip the balance permanently and irreparably in the direction of one reading only.

2.5. Orthographic jokes

The issue of orthographic jokes is similar to that of phonological jokes which has just been discussed insofar as they both concern the differences between jokes as to how they respond to the mode of communication. Not all jokes are equally affected by a switch from oral communication to written communication. By oral communication we mean the telling of a joke by means of spoken language. Written communication would include the more traditional sources, such as joke books, magazines, personal letters, murals, as well as the newer ones, such as television teletext and computer-mediated communication, particularly the Internet, where we have witnessed an explosion of Web sites holding humour tokens, a substantial proportion of them being jokes. The importance of the latter medium as an exchange platform for jokes has lately grown quite dramatically, and can be expected to keep growing with the continued expansion of the Information Superhighway. If only for this reason, the role of orthography in jokes deserve s more study. Essentially, the involvement of orthography cuts across the basic categorization, and it may be treated as an additional feature. It is of a continuous rather than discrete nature, and for this reason holds only weak taxonomic potential. At best, then, one can speak of a certain joke being "more orthographic" than another. For example, joke (24) represents the high end of the scale:

(24) When several telephone calls to a typewriter repair company failed to produce results, a man sat down and wrote the following letter:

Thmith Typewriter Thervice, Inc.

16 Joneth Thtreet

Harrithon, Miththiththippi

Thirth:

Will you kindly thend a man to my buthinethth addrethth to repair thith typewriter?

Thomebody broke the "eth" key.

Yourth thincerely,

Thamuel Thimpthon (Hoke 1965: 115)

It is inconceivable for anyone to try to actually tell joke (24), since spelling is essential to the joke. Incidentally, joke (24) can only be seen as based on ambiguity in a rather special sense: with the string th seen as being interpretable as a sequence of letters t+h and as one letter s.

Joke (25) below is less doubtfully ambiguity-based (on the other hand, its status as a joke may be in some doubt -- it derives from the well-known family of one-liner "do-it's"):

(25) Why do optometrists make efficient lovers?

Because they do it with the greatest of E's.

In (25) the spelling supports one interpretation which, when the joke is presented orally, will be missed by those not sufficiently familiar with the typical layout of a vision-test board. Therefore, the chances of getting (25) are better in the written mode of communication.

An example in which orthography plays a facilitating, but less crucial role than in (25) is presented in (26):

(26) A not so well-to-do family had to make a tremendous financial effort in order to send the oldest son to Oxford. The son settled in well, but after only two months he sent a letter home asking for 2CI0 pounds, so as he could buy a new suit for the freshmen's ball. After much discussion, the parents decided to send him the money. Two weeks Later they received another letter, saying:

"Thanks ever so much for the money. With my new suit on, everybody at the ball said I looked like a veritable count."

"Well," said the father, "but then again Tom never could spell."

Joke (26) is orthographic in the sense that it is more difficult to get without actually looking back at the words spelled out and recognizing that the word count has a second interpretation once bad spelling is factored in. Manipulating letters is easier when one can look at them. The operation of the recipient of this joke is to some extent the reverse of that of my wordprocessor's spellchecker, which gives count as the fourth suggested replacement for cunt, the latter being unknown to the spellchecker, but very well known to and easily retrievable by a normal speaker of English. Aside from the role orthography plays in (26), the joke should be placed in the lexical ambiguity category.

Looking back at joke (22) discussed in the previous section, it has to be noted that the potential variation in (suprasegmental) phonetic realization that has been suggested for (22) has no parallel in two different spelling representations. Quite the reverse is true of (23), where bean and been differ in the spelling. In this sense, (22) holds more clues distinguishing between the two interpretations in its spoken version, whereas (23) in its printed form.

2.6. Deictic reference

A further category of linguistic ambiguity jokes is based on deictic reference. Deictic elements rely heavily on context, and the set of their potential semantic referents is very extensive. The ambiguous string is in this case an anaphoric element that ultimately (in the mind of the joke recipient) refers to two distinct entities. Consider (27) below:

(27) Caller: "Hello? Maternity ward? I'd like to know whether Mrs. Jones, a patient of yours, has already delivered."

Receptionist: "Is this her first baby?"

Caller: "No, it's her husband."

The unit that is ambiguous here is the pronoun this. As it is a word-long unit, the joke could conceivably be classified as lexical. A deictic element, however, is special. Locating the reference for a deictic element would hardly involve retrieving one of a limited set of known senses that best fits the context, as would be the case for non-deictic lexical items. Rather, the context alone would be engaged. In (27), the receptionist uses this to refer to (the occassion of) the baby being born in the ward, while the caller interprets the word as referring to himself, in which case this would be used in the sense 'the person at the other end of the telephone line'. Substituting for the deictic element a formulation that allows only one of the two interpretations will remove the ambiguity. Thus, we could substitute either the person I'm speaking to or the baby being born. Both substitutions would yield parts of the exchange incongruous, precisely because those parts would only be compatible with the other interp retation which has been excluded by artificially removing the ambiguous element.

Joke (28) below is a somewhat unclear case:

(28) At a well-known religious college, the academics were excellent but the food was utterly monotonous. Practically every dinner it was tomato soup, creamed chicken, mashed potatoes, string beans, and ice cream with chocolate sauce. One student who took his religion seriously, but who could hardly stand the food, said the same grace every evening at dinner. He quoted from Hebrews, 13:8: "Jesus Christ, the same yesterday, and today, and forever." (Johnson 1989: 220)

The ambiguous string here is (the) same. As a deictic element, ii refers in this context anaphorically to Jesus Christ (the original interpretation of the quote), as well as to the menu at the college refectory. The ambiguity that results forms the backbone of joke (28). However, there is some doubt about whether in (28) the ambiguous string is a single lexical unit. The decision would depend on what is taken as a single lexical unit. Further complication is introduced by the fact that it is possible to analyze the ambiguity as arising from an elliptical construction, the choice being roughly between the same being and the same food. If this were the analysis, then the same would not so much refer to one or the other, as it would premodify a missing element. It would be the absence of the missing element that would be responsible for the ambiguity. It could be argued, then, that the ambiguity of (28) has some grounding in syntax. There is yet another sense in which syntax might come into play here. The syntac tic function of Jesus Chris appears to be different in the two readings. On the 'same being' reading, it could be viewed as appositive to the same, but not so on the 'same food' reading, where it is just an expletive, with a very loose syntactic connection to the material that follows it.

2.6.1. Deictic versus non-deictic interpretation

In joke (29) the ambiguity is between a deictic and non-deictic interpretation:

(29) "Which is the other side of the street?"

"Over there."

"That's strange, the gentleman over there said it was over here."

The enquirer interprets the other as a non-deictic reference, one that, in this case, is independent from the physical location of the speaker and hearer. The enquired and the "gentleman over there" interpret, we should assume in lack of evidence to the contrary, the other in the standard deictic manner. This disparity of interpretation by the different characters of the joke embodies the ambiguity of the joke. It is interesting to observe at this point that the enquirer does not seem to have any problems with some other deictic expressions, namely over there and over here, despite their similarity to the other. In the real world, such an inconsistency might be slightly unexpected, but in the world of joke characters it is quite usual (see Paulos 1977, 1980, 1985). Incidentally, the above observation appears to support the claim that the funniness of linguistic jokes stems largely from the local duality of interpretation (ambiguity), and not so much from the parody of a character. Should the latter be the cas e, we might expect a parody to make as much mileage out of a parodied feature as possible by making the character misinterpret over there in the same way as (s)he has done in the other case.

2.7. Specific versus non-specific interpretation

Another type of linguistic ambiguity on which jokes can turn is the possibility of specific and non-specific interpretation (as defined by Huddleston 1984 and 1988), typically of a noun phrase. On a non-specific interpretation, a noun phrase refers to any member of a class of possible referents. On a specific interpretation, it refers to a well-defined, delimited set of identifiable referents (often just one). A simple two-liner in (30) well illustrates this type of ambiguity:

(30) "In Los Angeles a man is hit by a car every five minutes."

"Boy, I'll bet he's pretty beat up."

The first statement of (30) as originally understood uses a man in a non-specific sense. The statement, on this reading, means approximately 'in Los Angeles pedestrians get hit by cars every five minutes (on average)'. The response contained in the second line of the joke gives away an alternative reading of the first statement, which is a specific reading along the lines of: 'there is a man in Los Angeles who gets hit by a car every five minutes'. It does so through the use of 'he', which forces the specific interpretation (except when used within one clause coreferentially with a non-specific noun phrase, but this is not the case here), and possibly also through saying that he's pretty beat up, which is consistent with being hit repetitively.

Joke (31) will serve as another illustration:

(31) The best man at a wedding replies to the bridegroom's mother's inquiry, "No, Mrs. Smith, I am not getting married any time soon. I would like to marry a woman who is rich, smart, and obedient." Another young man intervenes at this point, "May I have her number in the meantime?" (Raskin 1987: 20)

The best man in saying I would like to marry a woman who is rich, smart, and obedient intends to make a general statement about the kind of wife that he would prefer, should he choose to many. He does not point to any specific candidate, On the non-specific interpretation, even the presence of such a woman is not necessarily implied, which is more clearly appreciated if one considers statements such as: I would like a marry a woman who will never grow old. The young man appears to interpret a woman who is rich, smart, and obedient as a specific noun phrase, referring to one specific person. This is made evident in the punch line where he requests her telephone number. Only a specific person can have a telephone number that can be written down. Hence, an alternative reading of the noun phrase is presented to the recipient, making him or her aware of the ambiguity between the non-specific; and specific interpretation.

A final example of a joke based on the specific vs. non-specific type of ambiguity is (32) below:

(32) "I hear the bank is looking for a teller."

"I thought they hired one a month ago."

"That's the one they are looking for."

The original reading of the first line, which is consistent with the second line, informs the recipient that the bank is looking for some person to take up the position of a teller. The bank, on this reading, is looking for a non-specific person. The second line suggests that the bank should not have a need to look for a candidate, since the bank recently hired a new one, and thus the vacancy would have been filled. The third line introduces an alternative reading of the first line. The one can only be specific, so on this reading the bank is looking for a specific person: the newly employed teller who presumably stole some money from the bank and got away. At this point the recipient is aware that a teller in the first line of (32) can be a non-specific person, or a specific person, and thus there is ambiguity between the non-specific and specific reading. Incidentally, the speaker of the first line may not be obeying the Cooperative Principle of Conversation (Grice 1975), because he may not at this point be giving a sufficient amount of information. More about that in the following section.

2.8. Pragmatic ambiguity

Pragmatic ambiguity in jokes arises when the two interpretations of the ambiguous fragment are identical with respect to their syntactic structure, lexical content, and phonetic form, but different in terms of the pragmatic function which the fragment exhibits within the two interpretations.

By pragmatic function of an utterance I here mean the way in which the utterance affects or is intended to affect the state of mind of those involved, or the state of affairs. More specifically, it is the intention or result of extracting/giving specific information; the intention or result of producing certain verbal/nonverbal behaviour in some of the participants in the linguistic transaction; the intention or result of modifying the mental attitude towards, or perceived status of, an entity on the part of some of the participants. The discrimination between intention and result that I have repeatedly insisted upon is important in that it captures the ontological distinction and potential disparity between the conceived and the perceived. The ultimate goal of a bona-fide communicative interchange is to minimize this disparity.

In one type of joke this disparity is used as the source of ambiguity. Thus, it is exploited for the purpose of generating humour. Consider (33) below as an example.

(33) Two farmers had known each other all their lives, but their conversations were usually restricted to "Good morning" or "Nice day." One afternoon, however, the first farmer asked:

"Hi, Pete, what did you give your horse when he had the colic?"

"Turpentine," said Pete.

"Thanks," said his friend.

Two weeks later they met again.

"Didn't you tell me, Pete, that you gave your horse turpentine when he had the colic?"

"Yes," said Pete.

"Well, I gave mine turpentine and he died."

"So did mine," said Pete. (Misztal 1990: 506)

The ambiguity of (33) is contained in the first farmer's query (Hi, Pete, what did you give your horse when he had the colic?). The information that the first farmer wants to extract is how to treat his horse's colic. Stated literally, the request for information as intended by the first farmer might be something like 'based on your previous experience with horse's colic that I know of, what medication is effective in treating this condition?' It is only following the punch line (So did mine) that the recipient comes to realize that an alternative reading is possible of the first farmer's original query. This is a reading that Pete apparently adopts and it treats the query as an attempt at extracting information from Pete about what Pete specifically gave to his horse when it had the colic, regardless of what positive or negative effect it had on the horse's health. The exact scope of information that the first farmer wants to extract (as conceived by the first farmer) is different from that apparently percei ved by Pete. The word "apparently" is operative here, as two further possibilities arise here. Either Pete indeed perceives the first farmer's query in a more literal manner than is usual or else he perceives it the way the first farmer intended but pretends differently. In either case, the possibility of the alternative interpretation of the first fanner's original query is communicated to the recipient of the joke and thus generates ambiguity. In terms of Grice (1975; see also Pepicello 1987, Yamaguchi 1988, Attardo 1990), Pete violates the Maxim of Quantity (be as informative as required), by withholding a fact relevant and crucial to the first farmer's actual intention, the fact that the turpentine led to the death of Pete's horse. The violation can be seen as either unwitting or deliberate.

Another somewhat similar example is (34):

(34) "Jenny was washing the window on the fourteenth floor yesterday, and she fell off the window-sill."

"Good Lord, is she alive?"

'Yes. Fortunately, she fell inside."

This time it may again be claimed that crucial information is being withheld by the first character, but, more importantly perhaps, irrelevant information is given which then becomes misleading. The irrelevant piece of information is the floor number. Since Jenny fell inside, it does not matter which floor the window was on, as the distance between the window-sill and floor level is the same on every storey. Giving that information suggests that it is somehow relevant (if it is assumed that the participants respect Grice's Cooperative Principle of Conversation). The rather obvious scenario in which it would be relevant is if Jenny took a fourteen-floor fall, and this seems to be the conclusion drawn by the second character in the joke, since he or she expresses a serious concern for her being alive. The final line of the joke, containing the punch line, clarifies what really happened within the world of the joke and introduces the alternative interpretation of she fell off the window-sill: 'fell inside' rathe r than 'fell out the window'.

Speech Act Theory (Austin 1962, Searle 1969, Searle 1979, De Bruyn 1995) provides a possible framework for analyzing some pragmatic ambiguities, such as may be found in (35):

(35) Andy: "Have you forgotten you owe me five dollars for these ice-skates?"

Sandy: "No, but give me time and I will." (Hoke 1965: 16)

Andy does not really, at least not primarily, want to find out whether Sandy's memory still retains a record of the fact that he owes Andy money. Rather, Andy simply wants his money back, and his Have you forgotten you owe me five dollars for these ice-skates? can be seen as a request to Sandy to give him back the five dollars that Sandy owes him. In terms of felicity conditions, we may say that what Andy says probably isn't a felicitous question, as Andy doesn't probably care very much whether Sandy has forgotten about the debt, as long as he pays up. Sandy, on the face of it, interprets it as a direct question, more likely than not to wheedle out of returning the money, and in his response he only addresses the point of how well and for how long he will be able to retain the recollection of owing the money in his memory. The pragmatic function of Andy's utterance as intended by Andy is to make Sandy pay him back the money. The pragmatic function of the same utterance as ostensibly (though not actually) perc eived by Sandy is Andy's wanting to extract information from him about whether he still has a recollection of the debt. It is worth noting that although Sandy acts on the altemative reading of the ambiguous utterance, he is aware of the original reading, so he may indeed be said to perceive the ambiguity (and not just perceive: he uses it to his own advantage; in terms of Kopytko 1995, Sandy may be characterized by the pragmatic variable of Cunning). In a sense, then, the teller of jokes such as (35) (and also (5) above) is as if relating the joke which is originally told by one character of the joke. This is different from jokes such as (36) below:

(36) Departing Traveller: "Run upstairs, boy, and see if my umbrella is in room 483. I think I left it in the comer by the window." Boy (some minutes later): "Quite right, sir, it's exactly where you said it was." (Misztal 1990: 713)

In (36) the departing traveller wants to recover the umbrella that he accidentally left behind in the room. His intention in addressing the boy is to instruct him to fetch the umbrella from the room. The boy interprets this intended indirect request more directly and simply checks whether the umbrella really is at the place designated by the traveller, without fetching it. Incidentally, the boy could have been even more literal in interpreting the original request: overtly, there is nothing in the original request about coming back from the room in the first place, but the boy does that. Without this partial compliance with the intended force of the request, it would be difficult to make the joke work, since in the absence of a clear signal that the boy misunderstood the request, the audience would be likely to assume that some external circumstances, not communicated in the text, prevented the boy from returning from the hotel room, and the ambiguity of the request would not arise.

Another way of dealing with jokes (33), (35), and (36) would be to say that a joke character responding to an utterance made by another character fails to notice, or pretends not to notice, the pragmatic implications intended by the other character. In some other jokes, the situation may be reversed. Consider (37) below:

(37) Johnny, ten years old, applied for a summer job as a grocery boy. The grocer wanted a serious-minded lad for the job, so he put Johnny to a little test.

"Well, my boy, what would you do with a miliion dollars?" he asked.

"Oh, gee! I don't know. I wasn't expecting so much at the start" (Hoke 1965: 13)

The grocer intends the query as a direct question about what the boy would do if he happened to have a million dollars. The boy interprets this as an indirect offer of a million dollar pay. The boy reads pragmatic implications that were not intended by the grocer.

(38) is a slightly different case:

(38) Mike: "I wish I had the money to buy an elephant."

Spike: "Why? What would you do with an elephant?"

Mike: "Who wants an elephant? I just want the money." (Hoke 1965:13)

On the face of it, (38) looks quite similar to (37). Spike draws an inference that Mike wants to buy an elephant which Mike sets out to correct in his last line (the punch line disclosing the ambiguity of his original statement). However, whereas in (37) the grocer really does not intend to suggest he is offering the boy a million dollars (as explained by the narrator in the joke), in (38) it is likely that Mike indeed does intend to suggest the implication that he wants to buy an elephant, only later to pretend he didn't mean that. Seen in this light, joke (38) (along with (5) and (35) above) would be another example of "it's the character that's making the joke".

2.9. Type of modality 

Huddleston (1988: 78-79) recognizes three types of modality: epistemic, deontic, and subject-oriented. Briefly put, epistemic modality modifies the truth of a semantic proposition in terms of its probability (whether it's unlikely, possible, or probable); deontic modality involves the issuing of directives and is associated with notions such as permission or obligation; subject-oriented modality ascribes a certain property to the subject of a clause. As the grammatical category of mood is virtually nonexistent in contemporary English, certain lexical items may exhibit ambiguity between two different modality types. The short joke in (39) below displays an ambiguity between deontic (original interpretation corresponding to probable intention of the professor) and epistemic (alternative interpretation introduced by the student's response) modality:

(39) Professor: "You can't sleep in my class."

Student: "If you didn't talk so loud I could." (Misztal 1991): 143)

In turn, (40) exhibits the contrast between epistemic and subject-oriented modality:

(40) Stem librarian: "Please be quiet. The people near you can't read." Small boy: "Well, they ought to be ashamed of themselves! I've been able to read since I was six." (Hoke 1965: 6)

The librarian means can't read as 'find it difficult to read under the circumstances' (original interpretation, epistemic modality), whereas the boy understands can't read as 'do not possess the skill to read' (alternative interpretation, subject-oriented modality).

2.10. Cases of multiple ambiguity

Many jokes involve more than just a single ambiguity. With single ambiguity, a joke is constructed around an ambiguous fragment of text with two interpretations. The loci of ambiguity may be located on non-overlapping fragments of the joke's text (which more or less covers Hockett's (1973) complex and compound jokes), such as in (41) below:

(41) Pat bought a hot dog to guard the family valuables and it slept on the stove because it was a range rover. (Cagney 1979: 46)

or in Hockett's (1973: 155) example given in (42) below:

(42) Mr. Wong, a Canadian of Chinese extraction, visited the nursery in the maternity ward, and then hastened, much perturbed, to his wife's bedside. Said he: 'Two Wongs do not make a White!" Said she: "I can assure you it was purely occidental."

In both (41) and (42), the final parts (because it was a range rover and Said she: "I can assure you it was purely occidental." respectively) could be deleted without rendering the jokes unfunny. The deletion would eliminate some of the humorous potential of jokes (41) and (42), as ambiguous strings contributing to the funniness of original texts would no longer be present, but the texts would survive on the strength (though I only apply this word to (41) with hesitation) of the remaining ambiguities. In such cases I will speak of multiple independent ambiguity. Unfortunately, straightforward deletion of a fragment of text is not always a reliable test for independence of ambiguities. What we should more generally be looking at is whether an ambiguity can be eliminated without at the same time removing the other ambiguity (or ambiguities). For example, in (43) below recognizing one of the ambiguities assists in getting the other one:

(43) If you had three feet, what kind of work would you do? Yard work. (Klein 1992, no page)

Even so, there appear to be two independent ambiguities in (43): feet is ambiguous between 'lower extremities' and 'units of linear measurement', and yard can mean either 'outdoor area' or 'unit of linear measurement'. Indeed, it can be shown that the two ambiguities are independent by removing only one of them (though not through straighforward deletion), as in (44):

(44) If you were three feet tall, what kind of work would you do? Yard work.

We can also remove the yard ambiguity, as in (45), but some other (not necessarily ambiguous in itself) way of enhancing the 'unit of measure' reading should be provided:

(45) If you had three feet, what kind of work would you do?

I'd be a baseball player.

Why?

For a basketballer, three feet would be too short.

Another example of independent ambiguity of this kind is given in (46):

(46) Noah and his wife stood on the deck of the Ark and watched the flood receding.

"Right," said Noah, "now it's up to you to go forth and multiply."

All the animals left two by two, filing out of the Ark and disappearing into the distance.

"Well, that's it, dear," said Noah, sitting down for a cup of tea. Then, out of the corner of his eye, he spotted two snakes hiding beneath the kitchen table. "What are you two doing here?" he asked angrily, "I told you to go forth and multiply."

"We can't," hissed the snakes. "We're adders." (Davro 1987: 123)

On the other hand, there are jokes with multiple interdependent ambiguity, where the individual ambiguities cannot be isolated from one another. Often, we seem to be dealing with lexical plus syntactic class ambiguity, where the syntactic combinatory properties of a given lexical item prevent it from being freely combined with both syntactic representations of a phrasal construction of which the lexical item is a constituent, as in the fairly elaborate joke given in (47):

(47) Three small siblings had a pet sparrow, which, alas, died. The children were very sad, and they decided to give the dead bird a really good burial service. Their families were faithful members of the church, so the children had some ideas of how to go about it.

The first step was to dig the grave in a carefully chosen spot in a comer of the yard. Then they solemnly prepared for the actual interment. One child held the sparrow over the grave, and another recited, "In the name of the Father, and the Son, and in the hole he goes." (Johnson 1989: 262)

Thus, only two unitary interpretations can be maintained: 'in the hole he goes' and '(in) the Holy Ghost' despite the fact that we have two lexical ambiguities ('Ghost' vs. 'goes' and 'Holy' vs. 'hole he') and a syntactic class ambiguity (different phrasal structure: VP with a fronted PP constituent vs. PP with an NP constituent). This fact follows from the rampant ungrammaticality of strings like *in the hole he Ghost or *in the Holy goes.

At other times, the selectional restrictions may be more semantic in nature, such as in (48) below:

(48) What did the little termite in a pub say to his father?

Beat me, Daddy, I ate the bar. (Johnson 1989: 127, slightly modified)

Joke (48) only works for recipients who are familiar with the boogie-woogie piece "Beat Me, Daddy, Eight to the Bar" (see Raskin (1990) and Attardo (1994: 268) for the concept of sophisticated jokes). In this case, the original title song contrasts with the "as spelled" interpretation. As in (47) above, so too in (48) we have two lexical and one syntactic ambiguity, but again the VP interpretation 'I ate the bar' necessitates the choice of ate rather than eight, since the latter cannot head a VP (syntactic restriction). As for the choice between bar 'counter' and bar 'unit of timing in music', both would fit syntactically into the VP, but the verb to eat, when used of a furniture-consuming creature such as a termite, strongly favours the 'counter' interpretation, and so this preference is semantically motivated. Both (47) and (48), then, illustrate multiple interdependent ambiguity.

(1.) The present paper is partially based on Lew (1996a), where the interested reader will find more extensive exemplification as well as some additional categories of linguistic jokes.

(2.) See Lew (in press) for a justification and elucidation of this particular formulation.

(3.) Another difference in syntactic function between the two readings might be argued for, namely that between me as direct object versus indirect object in They left me an orphan. This would be far less convincing, however, because this is not an ambiguity in the actual clause as it stands in the joke, but rather in what might be taken as a pre-passivization form of this clause.

REFERENCES

Attardo, Salvatore

1990 "The violation of Grice's maxims in jokes", in: Kim Hall (ed.), 355-362.

1993 Review of Delia Chiaro. Humor: International Journal of Humor Research 6, 3:325-334.

Attardo, Salvatore -- Donalee Hughes Attardo -- Paul Baltes -- Mamie Jo Petray

1994 Linguistic theories of humor. Berlin, New York: Mouton de Gruyter.

1994 "The linear organization of jokes: Analysis of two thousand texts", Humor: International Journal of Humor Research 7, 1:27-54.

Attardo, Salvatore -- Victor Raskin 

1991 "Script theory revis(it)ed: Joke similarity and joke representation model", Humor: International Journal of Humor Research 4 (3/4): 293-347.

Austin, John 

1962 How to do things with words. Oxford: Clarendon Press.

Bader, Markus

1994 "Syntactic function ambiguities", Folia Linguistica 28 (1-2): 5-66.

Cagney, Peter (ed.)

1979 Official Irish joke book no. 4. London: Futura Publications.

Chapman, Anthony J. -- Hugh C. Foot (eds.)

1977 It's a funny thing, humour. International conference on humour and laughter. Oxford: Pergamon Press.

Chiaro, Delia

1992 The language of jokes: Analysing verbal play. London: Routledge.

Cieslicka-Ratajczak, Anna

forthcoming "Toward a model of idiom representation and processing", Studia Anglica Posnaniensia.

Cole, Peter -- Jeny L. Morgan (eds.)

1975 Syntax and semantics, Vol 3: Speech acts. New York: Academic Press.

1986 Intonation. Cambridge: CUP.

Cruttenden, Alan

Davro, Bobby

1987 Bobby Davro 's laugh-a-minute joke book. London: Knight Books.

De Bruyn, Pieter

1995 The role of indirect speech acts in the construction of jokes. [Paper presented at the 13th International Humor Conference, July 31 -- August 4, Birmingham, England.]

Fowles, Barbara -- Marcia E. Glanz

1977 "Competence and talent in verbal riddle comprehension", Journal of Child Language 4: 433-452.

Frumusani, Daniela

1986 "Context si conventie in producerea/receptarea glumei" [Context and convention in the production/reception of jokes], Studii si Cercetari Lingvistice 37,6: 509-516.

Goldstein, Jeffrey H. -- Paul Edward McGhee (eds.)

1972 The psychology of humor: Theoretical perspectives and empirical issues. New York: Academic Press.

Grice, H. Paul

1975 "Logic and conversation", in: Peter Cole -- Jerry L. Morgan (eds.), 41-59.

Hall, Kira (ed.)

1990 Proceedings of the 16th Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society.

Hirsh-Pasek, Kathy -- Lila R. Gleitman -- Henry Gleitman

1986 "What did the brain say to the mind? A study of the detection and report of ambiguity by young children", in: Anne Sinclair -- Robert J. Jarvella -- Willem J.M. Levelt (eds.), 97-132.

Hockett, Charles F.

1973 "Jokes", in: M. Estellie Smith (ed.), 153-178.

Hoke, Helen

1965 More jokes, New York: Franklin Watts.

Horgan, Dianne

1981 "Learning to tell jokes: A case study of metalinguistic abilities", Journal of Child Language 8,1: 217-224.

Huddleston, Rodney

1984 Introduction to the grammar of English. Cambridge: CUP.

1988 English grammar: An outline. Cambridge: CUP.

Hughes, Shaun F.D. -- Victor Raskin (eds.)

1990 Proceedings of the 1988 WHIM Conference held April 1-4, 1988 at Purdue University. West Lafayette, In, Tempe, Az.: ISHS, Purdue University.

Johnson, Eric W.

1989 Treasury of humor. Buffalo: Ivy Books.

Jucker, Andreas (ed.)

1995 Historical pragmatics. Amsterdam, Philadelphia: John Benjamins.

Lew, Robert

1996a An ambiguity-based theory of the linguistic verbal joke in English. [Unpublished Ph.D. dissertation, Adam Mickiewicz University.]

1996b "Exploitation of linguistic ambiguity in Polish and English verbal jokes", Papers and Studies in Contrastive Linguistics 31: 127-133.

in press "Ambiguity-generating devices in linguistic verbal jokes", Humor: International Journal of Humor Research.

Klein, Kathryn A.

1992 The effects of training on linguistic humor comprehension by 5th grade students. [Paper delivered at the annual session of Speech Science students at Eastern Michigan University on 11th December, 1992.]

Kopytko, Roman

1995 "Linguistic politeness strategies in Shakespeare's plays", in: Andreas Jucker (ed.), 515-540.

Makkai, Adam

1972 Idiom structure in English. The Hague: Mouton.

McArthur, Tom (ed.)

1992 The Oxford companion to the English language. Oxford, New York: OUP.

McGhee, Paul Edward -- Anthony J. Chapman (eds.)

1980 Children's humour. Chichester John Wiley and Sons.

Milner, George Bertram

1972 "Homo ridens: Towards a semiotic theory of humour and laughter", Semiotica 5: 1-30.

Misztal, Mariusz

1990 Life in jokes. Warszawa: Wydawnictwa Szkolne i Pedagogiczne.

Paulos, John Allen 

1977 "The logic of humour and the humour of logic", in: Anthony J. Chapman -- Hugh C. Foot (eds.), 113-114.

1980 Mathematics and humor. Chicago and London: University of Chicago Press.

1985 I think, therefore I laugh: An alternative approach to philosophy. New York: Columbia University Press.

Pepicello, William J.

1987 "Pragmatics of humorous language", International Journal of the Sociology of Language 65: 27-35.

Pocheptsov, Georgij

1974 Language and humour. Kiev: Vysca Skola Publishers.

Raskin, Victor

1985 Semantic mechanisms of humor (Synthese Language Library 24.) Dordrecht: D. Reidel.

1987 "Linguistic heuristics of humor. A script-based semantic approach", International Journal of the Sociology of Language 65: 11-25.

1990 "Sophisticated jokes", in: Shaun F.D. Hughes -- Victor Raskin (eds.), 125-127.

Searle, John R.

1969 Speech acts. An essay in the philosophy of language. Cambridge: CUP.

1979 Expression of meaning: Studies in the theory of speech acts. Cambridge: CUP.

Shultz, Thomas R. - Frances Horibe

1974 "Development of the appreciation of verbal jokes", Developmental Psychology 10,1: 13-20.

Shultz, Thomas R. - Robert Pilon

1973 "Development of the ability to detect linguistic ambiguity", Child Development 44: 728-733.

Shultz, Thomas R. - Judith Robillard

1980 "The development of linguistic humour in children", in: Paul Edward McGhee - Anthony J. Chapman (eds.), 59-90.

Sinclair, Anne - Robert J. Jarvella - Willem J.M. Levelt (eds.)

1986 The child's conception of language. Berlin: Springer Verlag.

Skowronska, Natalia

1989 English and Polish linguistic jokes: A contrastive study. [Unpublished MA dissertation, Adam Mickiewicz University.]

Smith, M. Estellie (ed.)

1973 Studies in linguistics in honor of George L. Trager. The Hague: Mouton.

Sobkowiak, Wlodzimierz

1991 Metaphonology of English paronomasic puns. Frankfurt am Main: Peter Lang.

Spector, Cecile C.

1990 "Linguistic humor comprehension of normal and language-impaired adolescents", Journal of Speech and Hearing Disorders 55: 533-541.

Stanley, Richard

1980 The end. Harmondsworth: Puffin Books.

Yamaguchi, Haruhiko

1988 "How to pull strings with words", Journal of Pragmatics 12,3: 323-337.