Consequences

According to the proposed convention, there are constraints on the kind of situation that an innovative denominal verb may denote. It has to be (a) the kind of situation (b) that the speaker has good reason to believe (c) that on this occasion the listener can readily compute (d) uniquely (e) on the basis of their mutual knowledge (f) in such a way that it encompasses the parent noun and the other surface arguments of the verb. These constraints tell us not only what a verb will be taken to mean on particular occasions, but also when and why it will be judged acceptable or unacceptable. These constraints interact, and so are difficult to examine separately. Instead, we will consider seven major consequences of their interaction.

4.1. mutual knowledge. The kind of situation that a verb denotes, according to condition (e) above, is intended to be computed on the basis of things mutually known or believed by the speaker and listener. Normally this constraint is easy to satisfy, since most of the needed facts belong to the core of generic knowledge; and these core facts are ordinarily assumed to be mutual knowledge. When two strangers meet, they normally assume that each knows — and that each knows that the other knows, etc. — almost everything in this core. So when they want to create a new verb interpretable to everyone, all they need do is make sure it is computable from the facts in this core. This is a property of most denominal verbs, both the well-established and the innovative.

How do listeners decide which kind of situation a verb picks out? Consider how they see that brick in brick the ice cream means ‘form into the shape and size of a brick’. By conditions (a) — (d), the verb must denote ‘the kind of situation which the speaker has good reason to believe that on this occasion the listener can readily compute uniquely’. The speaker would have just such a reason if the kind of situation denoted were a salient, or conspicuously unique, part of core knowledge (see Lewis). But the most salient part of one’s generic knowledge of bricks is their predominant features — e. g., that they have a rectangular shape and child’s-shoebox size. Because of this salience, the listener can readily infer that the speaker could well have intended brick to denote a kind of state, event, or process having to do with these predominant features. The choice of this kind of situation, of course, depends on the other constraints as well; but the salience of these predominant features is critical.

The denominal verbs that we classified earlier provide excellent evidence that the salience of predominant features is truly critical. Most of these verbs are well established. For them to have become well established, when they were created, they had to have been interpretable to nearly everyone. And for that to have happened, most of them had to havebeen computable from the core of generic knowledge — in particular, from the salient parts of this core. Most of these verbs, then, should reflect the predominant features of the entities denoted by their parent nouns; and they do. As we noted earlier, when these concrete entities are classified by their predominant features, they fall into such classes as placeables, places, and agents. These classes correspond almost exactly to such classes among the denominal verbs as locatum, location, and agent verbs. So the very classification provided earlier is evidence for the use of constraints (b) — (e).

Many generic theories of concrete objects, however, have, more than one predominant feature; and so the corresponding well-established denominal verbs are often ambiguous. Shelves, for example, have at least two predominant features: they are places that things are put on, and placeables that are put on walls. This has allowed shelf toestablish two meanings, those in shelve the books and shelve the closet. Other well-established verbs with at least two interpretations include floor the rooms / opponents; lime the walls / starlings; riddle the potatoes / the door with holes; brick the fireplace / cheese; curb the dog / street; cream the butter / coffee; tree the cat / avenue; and powder the nose / aspirin. Listeners resolve these ambiguities, as usual, by selecting the interpretation that theybelieve the speaker could reasonably expect them to pick out uniquely on that occasion. They can often do this merely by consulting the direct object of the verb: shelve the books means ‘put the books on shelves’ because it couldn’t reasonably mean ‘put shelves on the books’.

Some concrete objects have predominant features that lead to a remarkable type of ambiguity. Two predominant features in the generic theory for ‘milk’, for example, are that milk is a substance put into or onto certain foods (its potential roles) and that it is a substance extracted from the mammary glands (its ontogeny). Consequently, milk has developed two meanings. In milk the tea it means ‘put milk in’; in milk the cow it means ‘take milk out’. This type of ambiguity is remarkable because the second interpretation is contradictory to the first — which in List 1 we labeled ‘in’ and ‘not-in’. Other verbs that have developed contradictory senses include seed the lawn / grapes; scale his hand / fish; cork the bottle / oaks; wind the organ / man; fleece the stones with moss / the sheep; top the cake / tree; shell the roadbed / peanuts; fin the boat / fish; girdle the waist / tree; the tree barked over / bark the tree; and dust the shelf.

It is easy to see how such contradictory interpretations can arise. Negative locatum verbs generally have parent nouns that denote parts of whole objects. Rinds, fleeces, and cores are proper parts of lemons, sheep, and apples (see Brown). Normally, these parts can be moved in only one direction with respect to their wholes — out or off — and so rind the lemon, fleece the sheep, and core the apple are all interpreted negatively. Yet the objects denoted by some of these parent nouns are conventionally placed in or on something else — as milk goes into tea or onto cereal — and this leads to the positive, the contradictory, interpretations in milk the tea and milk the cereal. The contradictory interpretations appear able to develop because they arise in contexts where they would not be confused[2].

However, some kinds of situations can be made salient in context, even though they are not based on predominant features, as in celluloid the door open, hairpin the lock open, bottle the demonstrator, and pie the woman. Thus hairpin, by virtue of its reduced complement the lock open, must denote a kind of situation in which a hairpin is an instrument for opening the lock. This rules out any predominant feature of hairpins, which are normally instruments for pinning up hair. The listener is expected to consult his knowledge of the physicalproperties of locks and hairpins, and infer that the hairpin is to be used as a picklock. The salience of the necessary mutual knowledge may be only momentary. Pie, for example, was interpretable by Time readers as ‘throw a pie in the face of’ only because it appeared in a story about pie-throwing.

Although most kinds of situations are made salient through allusions to generic knowledge, some are made salient through allusions to mutual knowledge of particulars. Thus the sense of My sister Houdin’d her way out of the locked closet relies on the speaker’s and listener’s mutual knowledge of a particular historical figure, Houdini, and of particular events in his life. The same goes for all other innovative verbs built on proper nouns.

Mutual knowledge of particulars is also critical for many innovations built on common nouns. The intended denotation of Max teapotted the policeman was salient in context only because of Ed’s and Joe’s mutual knowledge about Max and his compulsion. When the circumstances are just right, such particular knowledge may even override otherwise plausible generic knowledge. Normally, closet in Professor Jones closeted three students last week would have the sense ‘put and keep in a closet’. But for the speaker and listener who mutually knew that Jones had a pillory in her closet for punishing students — a piece of particular knowledge — closet would take on the sense ‘punish by means of the pillory in Jones’ closet’. Instances of this kind are not all that rare.

Mutual knowledge is also critical for the interpretation of verbs as distributive or collective. By virtue of generic knowledge, stamp the envelopes would normally be construed to mean that there was one or more stamps per envelope (a distributive interpretation), not that there was one or more stamps for the envelopes taken as a set (a collective interpretation). Blanket the children, however, could be taken either way; e. g., unlike stamp, it could be collective, with one or more blankets for the children as a set, if there were two children in one bed. Mutual knowledge of the particulars in each context is normally required to decide the issue.

This issue, however, tends to have a uniform resolution for location verbs with mass or collective parent nouns. Note that shellac the wall requires shellac to cover the wall: a dab of shellac on the wall won’t do. Given the uniqueness condition (d), this makes good sense. The speaker must have good grounds for believing that the kind of situation denoted by shellac can be computed uniquely. But how much of the wall is covered by shellac? The only unique but reasonable answer is ‘as completely as would be expected under the circumstances’. Connie powdered her nose, therefore, implies that the powder covered all those parts of Connie’s nose that the listener would expect to be covered. Its precise interpretation depends on the speaker’s and listener’s assessment of their mutual knowledge about nose powdering, Connie, and her habits. And carpet the floor, carpet the room, and carpet the house would ordinarily all be construed to mean that the carpet went only on the floors — consistent with our generic theory for carpets — and, for carpet the house, only on those floors that are normally expected to be covered (which excludes the kitchen, the cellar, and the garage). Yet carpet the wall would not be construed to mean that the carpets went on the floor, since wall contrasts with floor, and rules out just such an interpretation. The point of the examples in this section is that mutual knowledge is essential to the interpretation of innovative denominal verbs. To select the unique sense intended, on a particular occasion, the listener must decide which of the possible senses is most salient. Generally he can look to the predominant features of the generic theory associated with the parent noun, which will always be fairly salient. But salience is a relative notion, and depends on context. The listener must always assess his and the speaker’s mutual knowledge of the particulars in the present context, since that may make some other sense, the most salient. The listener’s ultimate goal is to find that sense which ‘the speaker has good reason to believe that on this occasion the listener can readily compute uniquely’.

4.2. kinds of situations. According to condition (a), an innovative denominal verb is intended to denote a kind of situation. What is meant by ‘kind’? That is a difficult epistemological question that we would not presume to answer here; but to get off the ground, let us begin with an intuitive characterization. A kind is a class or category of things with a rationale for membership, based on human conceptual and perceptual principles. So a kind is not a category of arbitrarily chosen things, like the class consisting of male humans and pine needles, but a category with a rationale that makes sense, like the class consisting of male humans and female humans.

Rationales for kinds come in many different forms. One we have already mentioned is the optimization of cue validity: things in the world tend to be categorized into kinds such that things within each category are as similar as possible to each other, and as different as possible from things in other categories. It is on this basis that fruits divide into such kinds as apples, oranges, and bananas; that furniture divides into such kinds as tables, chairs, and bookcases; and so on. Things can also be categorized into kinds on the basis of just one or a few distinguishing properties; thus dogs, cats, turtles, and goldfish constitute a kind because they are domesticable.

Some rationales for kinds, however, are apparently valued more highly than others; and the kinds they define are therefore deemed ‘better’ than others. For example, rationales based on permanent, inherent properties seem more highly valued than those based on temporary, non-inherent ones. ‘Round-things’ make а better kind than do ‘things likely to be found in a garage’. And rationales based on prominent properties seem to be more highly valued than those based on non-prominent ones. ‘Foods that taste sweet’ make a better kind than ‘foods that are rich in calcium’. In effect, kinds lie on a continuum from those with highly valued and obvious rationales to those that are so arbitrary that any rationale they can be given will seem ad-hoc.

When it comes to situations, kinds are particularly difficult to characterize, because situations are themselves difficult to characterize. Unlike concrete objects, situations do not come ready-made in discrete bundles. An act of sauté ing for instance, has no clear beginning or end; it may or may not include fetching the butter, turning on the heat, and scrubbing the pan afterward. Like other situations, it has vague boundaries. Situations are also vague in their range. Cooking may include sauté ing, frying, and broiling; but what about roasting marshmallows, making popcorn, and defrosting orange juice? Vagueness in boundaries and range is typical of most kinds of situations.

Yet situations clearly fall into kinds when they have good rationales. One common rationale is that a kind of situation consists of all situations that have the same goal, purpose, or outcome. Thus sauté ing consists of activities whose goal is the irreversible change in food to make it suitable for eating. Another common rationale is that a kind consists of all situations that employ the same means toward some end. Sauté ing, to continue the example, consists of activities whose means requires the application of heat in a pan with hot fat. For sauté ing, in fact, both the goal and the means seem to be critical. It is easy to see how, with these rationales, sauté ing would have both vague boundaries and a vague range. But these examples give only a flavor of possible rationales for kinds of situations. A full account of innovative denominal verbs will require a proper epistemological theory of situations and how they fall into kinds.

Kinds are critical to the interpretation of innovative denominal verbs. For any such verb, there may always be a unique class of situations that the verb could denote in that context; but uniqueness isn’t enough. Unless this class constitutes a kind with a highly-valued rationale, the verb will be judged relatively unacceptable in context. In He Houdini’d something, for example, Houdini could denote the unique class of things that we know Houdini did — escape from locked boxes, dote on his mother, and unmask fraudulent mediums. But this class is too diverse to have a highly-valued rationale, and so in this instance the verb is not very acceptable. But change the sentence to He Houdini’d the locks open or He Houdini’d his mother or He Houdini’d the fake palm reader, and suddenly it becomes more acceptable. Now the unique class that one can arrive at has a highly-valued rationale, since any one of these activities — picking a lock, doting on one’s mother, or unmasking frauds — is a kind with a common means, a common end, or both. As another example, She wanted to Richard Nixon her friend allows us to arrive at a unique class of situations — the class of things Richard Nixon did that one person could do to another. This diverse class, however, does not have a rationale that is highly-valued; hence Richard Nixon as a verb is not very acceptable. Yet change the sentence to She wanted to Richard Nixon a tape of the damaging conversation she had had with her friend, and it suddenly becomes more acceptable. Now the class of situations has a common means and a common end — erasure to get rid of incriminating evidence — and that makes it a better kind.

4.3. specificity. Another consequence of the convention on innovations is captured thus:

(26) Principle of specificity: the kind of situation that an innovation denotes is intended to be as specific as the circumstances warrant.

On hearing Margaret jetted to London, for example, we are warranted in inferring that she traveled by jet, but not that she travelled by just any type of airplane (which is not as specific as the circumstances warrant), or that she traveled by 747 (which is more specific than the circumstances warrant). The rationale for this principle is straightforward. According to condition (f), the kind of situation denoted by jetted must be one in which jets, Margaret, and London play roles. For this kind of situation to be unique — condition (d) — the means of transportation could not be anything more general than jets. We would have no way of deciding uniquely among the possibilities: any type of fast airplane, any type of airplane, any type of flying machine, or any type of vehicle. Nor could we infer a means of transportation that is more specific (like a 747, 707, or Concorde), because we could still not do so uniquely. It must be just right — as specific as the circumstances warrant.

With many denominal verbs, the more specific the circumstances, the more interpretable the verb becomes. In He Houdini’d something, the context warrants a kind of situation that is so general and diverse that it doesn’t have a very highly-valued rationale. With increasing amounts of syntactic context — by condition (f) — Houdini becomes more and more interpretable, as in this sequense: He Houdini’d the locks; He Houdini’d the locks open; He Houdini’d the box’s locks open; and He Houdini’d the box’s locks open from the inside. With each additional restriction, the rationale for the kind that one can infer becomes more highly valued. There is a similar progression in Abner spatulaed; Abner spatulaed the pancakes; Abner spatulaed the pancakes over; and Abner spatulaed the pancakes over with a flick of his wrist. Greater specificity alone, of course, isn’t sufficient for better interpretations, but it often helps.

Yet there must always be a reason for greater specificity as warranted by circumstances. This is particularly evident in the choice of parent nouns. In selecting Margaret 747’d to London over Margaret jetted (or flew) to London, the speaker must want to contrast flying by 747 with other modes of flying. This kind of contrast is not always appropriate. For example, in most circumstances Julia Chevied downtown would be too specific. But if a speaker wanted to contrast Julia’s use of the Chevy with her use of another car, he might say Julia didn’t take the saab downtown – she chevied there, with stress on Saab and Chevy to mark the contrast. As it happens, distinctions this specific are rarely needed; and so denominal verbs, especially the well-established ones, tend to be at the so-called ‘basic’ or ‘generic’ levels of abstraction — the level generally used for naming concrete objects.

4.4. pre-emption. Another consequence of conditions (b) through (d) is embodied in the following principle:

(27) The principle of pre-emption by synonymy: if a potential innovative denominal verb would be precisely synonymous with a well-established verb, the innovative verb is normally pre-empted by the well-established verb, and is therefore considered unacceptable.

The rationale of this principle can be illustrated for hospital, an innovative verb intended to mean ‘put into a hospital’. By conditions (b) — (d), the speaker must have good reason to believe that the listener can readily compute the intended sense uniquely. Thus the listener would ‘reason’ as follows: suppose my interlocutor had intended to convey the sense ‘put into a hospital’. If he had, he would have used the well-established verb hospitalize, which means precisely ‘put into a hospital’, because then he would have had good reason to think I would compute the intended sense uniquely. Since he used hospital, he must have meant something distinct from ‘put into a hospital’. Yet the only reasonable sense I can come up with is ‘put into a hospital’, which I already know to be impossible. Thus I find hospital to be uninterpretable, and therefore unacceptable.

There appear to be several distinguishable bounces of pre-emption by synonymy. The three we have identified are suppletion, entrenchment, and ancestry.

4.4.1. suppletion. In the paradigm of work / worked, talk / talked, jog / jogged, etc., the past tense of go is not goed but went. This is an instance of suppletion: goed is unacceptable because of the presence of the suppletive form went. There is a similar type of suppletion among denominal verbs. The regular way of forming a verb meaning ‘go by [vehicle]’ is to make a denominal verb from the unadorned name of the vehicle, as the noun verb paradigm helicopter / helicopter, bicycle / bicycle, taxi / taxi, canoe / canoe, etc. But there are two striking gaps in this list, car / car and airplane / airplane; i. e., Jack carred downtown and Connie airplaned to London are unacceptable. They appear to be ruled out because of the presence of the suppletive forms drive and fly, which in the same contexts would mean precisely ‘go by car’ and ‘go by airplane’. Just as go / goed and bring / bringed are replaced in the past-tense paradigm by go / went and bring / brought, so car / car and airplane / airplane are replaced in the denominal-verb paradigm by car / drive and airplane / fly.

Car and airplane are not pre-empted because the names of cars and airplanes in general cannot surface as verbs. In the right circumstances, Julia Chevied downtown and Connie 747’d to London are quite all right. Indeed, motor and auto are two relatively obsolete denominal verbs that actually do mean ‘go by car’. According to the OED, these appeared in 1896 and 1898 — when, of course, drive still meant ‘go by horse-drawn vehicle’, and therefore contrasted with motor and auto. With the passing of horse-drawn vehicles, drive has come to mean ‘go by car’, and now pre-empts the verb car, whose parent noun is the ordinary name for this vehicle. The verbs motor and auto, based on uncommon names for car, have been retained — but with a quaint, dated flavor for a specialized register that makes each of them contrast in meaning with drive.

Suppletion can also be exemplified among the terms for body parts. Thus elbow, hip, shoulder, finger, eye, chin, and other body-part names occur as denominal verbs; but fist the man in the face, palm his face, foot him in the knee, lip someone on the cheeks, and fingernail his back do not — at least in the senses of ‘hit’, ‘slap’, ‘kick’, ‘kiss’, and ‘scratch’. The reason, we suggest, is that they are pre-empted by these well-established verbs with which they would be precisely synonymous. Of course, the expressions with fist, palm, foot, lip, and fingernail can be used innovatively; but then they are taken to mean something different. Thus fist the man in the face could mean ‘grind a fist into the face of the man’, and palm his face could mean ‘brush his face with one’s palm’; but they could not mean simply ‘hit’ or ‘slap’. On the other side of the coin, the verbs eye and eyeball, which at first seem to be synonymous with look, actually belong to the large semantic field of looking terms — look, watch, observe, view, ogle, survey, regard, gawk at, stare at, etc. — and form subtle contrasts with each of them. Indeed, although foot the ball through the uprights is sometimes used by sportswriters to mean something like ‘kick’, it is always intended to contrast with kick in some way or another — in register or humor. So suppletion appears to account for the unacceptability of certain denominal verbs, and for the contrast in meaning in others.

4.4.2. entrenchment. In pre-emption by entrenchment, the presence of one idiomatic denominal verb prevents the formation from the same parent noun of a second denominal verb with the same meaning. Hospitalize, built on the noun hospital, is so entrenched that it pre-empts the creation of the second verb hospital with the same meaning. There are many such cases in English. Prison the thief, parallel to jail the thief, is pre-empted by the well-entrenched imprison the thief; tomb is pre-empted by entomb, pollen by pollenate, and throne by enthrone. When there are two denominal verbs formed from the same parent noun, they contrast in meaning, e. g. winter vs. winterize and list vs. enlist. In other words, if their meanings would be identical, the entrenched verb always takes precedence over and pre-empts the newcomer.

4.4.3. ancestry. Some denominal verbs are pre-empted because the parent nouns are themselves formed from verbs that are synonymous with their grandchildren. Thus, while butcher the meat is acceptable, baker the bread is not. To baker appears to be pre-empted by its obvious ancestor, bake, with which it would be synonymous. To butcher is acceptable because it has no such ancestor. Pre-emp­tion by ancestry also seems to account for the unacceptability of to farmer the hillside, to banker the money, and to driver the car, which are otherwise similar to to umpire the game, to volunteer the information, and to chauffeur the car. As before, however, a denominal verb can be acceptable if it contrasts in meaning with its grandparent. Sweeper the floor is acceptable, despite the presence of sweep, because sweeper entails the use of a carpet-sweeper, while sweep does not. An obvious ancestor, therefore, will pre-empt its descendant denominal verb if its descendant would have the identical meaning.

These three types of pre-emption — suppletion, entrenchment, and ancestry — prevent true synonyms. The obvious verb senses of car, hospital, and baker are prevented outright by drive, hospitalize, and bake; while the senses of palm, list, and sweeper are forced to be distinct from those of slap, enlist, and sweep. As Bolinger and others have argued, language in general eschews complete synonyms. This tendency may reflect the general applicability of conditions (b) — (d). When an expression has a true synonym, the speaker must have a good reason for selecting it over its alternative; and the listener, to satisfy unique computability, will try to find one. This does two things. It prevents the speaker from creating new expressions that are completely synonymous with old ones; and it forces him to add distinctions whenever he uses one of two expressions that would otherwise be completely synonymous. Both forces tend to prevent the creation of true synonyms.

A fourth type of pre-emption works by homonymy instead of synonymy, as stated in this principle:

(28) Principle of pre-emption by homonymy: if a potential innovative denominal verb is homonymous with a well-established verb and could be confused with it, the innovative verb is normally pre-empted, and therefore is considered unacceptable.

So Jan Dodged to New York, meaning ‘Jan went to New York by a Dodge’, is normally unacceptable, because it would be confused with the common verb meaning ‘shift suddenly’. The same goes for Jan Forded to New York, though the parallel construction Jan Chevied to New York is quite all right in the appropriate contrastive context. To take another semantic domain: to summer, autumn, and winter in France is acceptable, but to spring and fall in France is not, being pre-empted by the homonymous common verbs spring and fall. When marked for past tense, as in She springed and falled in France, they get even worse: the speaker now sounds as if he had also added the wrong inflections. Homonymy, then, is still another source of pre-emption.

4.5. ready computability. Condition (c) on its own requires that the sense of a denominal verb be one that the listener can compute readily. The idea is that, although some denominal verbs may be comprehensible on every other count, they may not be readily computable in this context by this particular listener. As an analogy, imagine that Helen told Sam, I asked Linda and Winifred over tonight, but the older one couldn’t make it. If Linda and Winifred were of similar age, and Helen knew that Sam couldn’t figure out which was older without considerable thought, she shouldn’t have referred to Winifred as the older one. Its referent, although computable, cannot be computed readily. She could have used the latter instead, since its referent is computed readily from information easily accessible in what she said. The relative acceptability of the older one and the latter, then, depends not merely on computability, but on ready computability. These examples have their parallels in innovative denominal verbs.

In order for the sense of an innovation to be computed readily, the listener must ordinarily be able to bring to mind very quickly the information necessary for its computation. Many people, for example, would find It’s stratusing right now, uttered by the first person they met in the morning, to be unacceptable — even though, with some thought, they could figure out what it meant. But said by a television weather reporter, in a discussion about cloud formation (as it actually was), it is quite acceptable. In that context, the listener finds it easy to recall that stratus is a type of cloud, and to see that the speaker must be talking about cloud formation. Similarly, My telephone was Hoovered once would be unacceptable in many contexts, since it would be difficult to discover the eponym J. Edgar Hoover and to sift through all we know about his activities to arrive at wiretapping. But in a conversation about FBI wiretapping, an utterance of the same sentence is fully acceptable. Accessible information seems to be crucial to ready computability.

Accessibility of information can also make a difference to what a verb is taken to mean, since one criterion for the salience of a particular kind of situation is its accessibility in memory. Thus, in Roger speared a cake of soap, spear would ordinarily be taken in one of its conventional senses, ‘pierce as with a spear’. But in a conversation on how Roger had managed to carve soap into different shapes for a display on hunting, it would be taken as an innovative denominal verb meaning ‘form into the shape of a spear’. In this context, the conventional meaning of spear is pre-empted, not merely because Roger’s carving activities are mutually known by the speaker and listener, but also because that knowledge is readily accessible. The speaker can be confident the listener will see this information to be relevant, simply because it is so accessible in this context. So the constraint on ready computability is an important part of the innovative denominal verb convention.

4.6. rhetorical considerations. Why invent denominal verbs? The main reason, perhaps, is economy of expression. As Grice notes, the speaker who observes the cooperative principle will try to avoid unnecessary prolixity. When someone says I guitared my way across the US, he is trying to pack into guitar what would otherwise have taken him many words to express. This economy is especially useful in new areas of technology for which there are too few verbs for situations that occur constantly. In computer circles, for example, people have evolved such denominal verbs as key in the data, flowchart the program, program the system, output the results, and CRT the trace, along with many others that are utterly opaque to outsiders. New technologies seem to be responsible for many of the denominal verbs that are now very common — Xerox, telephone, wire, radio, and paperclip. In each case, a complicated situation is expressed economically in a single verb.

Economy of expression apparently has its rewards. First, there is precision. For the hospital worker, autoclave the scalpels is more precise than sterilize the scalpels, and yet takes no longer to say. Second, there is vividness. For a political writer, it is more effective, to say The mayor tried to Richard Nixon the tapes of the meeting than to use erase in place of Nixon. The allusion to Nixon calls forth an image of an unscrupulous politician trying desperately to cover his tracks — an image that even a longer description could never capture adequately. There seems to be an intrinsic value to making allusions without belaboring them. Third, there is surprise. Jokes, witticisms, and other rhetorical devices depend for their effect on surprise, which in turn depends on economy of expression. This effect is exploited daily by such newspaper columnists as Herb Caen: The SF progress is not a biweekly, as erratum’d here yesterday, but a semi-weekly, and Chevy [Chase] especially has been chop-sticking all over the place, starting with Kan’s.

When economy of expression is taken too far, it loses its ready computability, and the result is inelegant. Some verbs seem inelegant because they are cumbersome, as in We Fourth-of-July’d at Lake Tahoe. This inelegance, however, is sometimes used deliberately, for comic effect, as in Punch’s He extract-of-beefed his bread. Other verbs require so much extra work in computing that the effort doesn’t seem worth it. While the attested example Karen weekended in the country seems good enough, Karen Saturdayed in the country does not. It appears that the effort demanded for computing Saturdayed outweighs any economy of expression, although this too may be an asset for comic effects.

There are also some clear cases of morphological confusion, where an innovation is unacceptable because its parent noun is already inflected for tense or number. Compare these forms:

(29) *John United’d / United-Airlines’d / Trailways’d to Los Angeles.

(30) John UA’d / American’d / Greyhounded / Air-California’d / Hughes-Air-wested / PSA’d to Los Angeles.

In 29 the verbs are cumbersome: they seem difficult to parse, and their parent nouns difficult to identify. It isn’t just the double inflections in 29 that make them awkward, for they are equally unacceptable in infinitival constructions:

(31) *John decided to United / United-Airlines / Trailways to Los Angeles.

In these verbs, the inflections on the parent nouns seem to conflict with construing the verbs as infinitives. Note, however, that the adjective-forming suffix in American in 30 does not lead to such unacceptability. In general, then, parent nouns ending in -ed or - s lead to morphologically confusing innovations, avoided because they are not so readily computable.

4.7. syntactic constraints. By condition (f), the kind of situation that a verb denotes must encompass the parent noun plus all the verb’s surface arguments. For Julia centrifuged the solution, the kind of situation denoted must simultaneously involve Julia, centrifuges, and the solution — not just Julia and centrifuges, or just centrifuges and the solution. That can be done if Julia is an agent, the solution is the patient of her action, and the centrifuge is the instrument by which her action is carried out. This constraint, along with conditions (a) — (e), is fulfilled if centrifuge is taken to mean ‘separate by means of a centrifuge’. The kind of situation denoted by a verb, then, will change with the surface arguments present. Tent has different meanings in David tented the blanket; David tented the baby before the storm hit; The marines tented the hillside; and David tented near the river, each depending on the surface arguments present. This condition, of course, has played a critical role in our discussions of mutual knowledge, kinds of situations, specificity, and ready computability.

Many innovative verbs require more inferential filling in than either centrifuge or tent, because the parent noun does not play such a direct role in the kind of situation denoted. Consider these five uses of siren (34 – 35 are examples we have actually heard):

(32) The fire stations sirened throughout the raid.

(33) The factory sirened midday and everyone stopped for lunch.

(34) The police sirened the Porsche to a stop.

(35) The police car sirened up to the accident.

(36) The police car sirened the daylights out of me.

In 32 the siren’s role is already indirect, since siren means ‘produce a wailing sound by means of a siren’; and in 33 – 36, it is this sound that is critical. In 33 – 34, the sound is used as a signal — for midday in 33, and for the Porsche to stop in 34 — but the way it works in the two instances is distinctly different. In 33 it is a time marker, and in 34 a police warning; these two aspects are meant to be taken as part of the situations denoted. In 35, the siren’s role is still less direct. To account for the police car doing something involving a siren up to the accident, one is led to the sense ‘drive quickly accompanied by the sound produced by a siren’. The warning function of the siren here is less central. Example 36 is particularly interesting, for it is a syntactic blend of siren and the idiom scare the daylights out of. The superficial arguments the daylights and (out of) me together signal the presence of the idiom, and siren itself requires that this scare involve a siren. One is therefore led to the sense ‘scare by means of the sound produced by a siren’. Syntactic blends of this sort are not uncommon.

This sketch of syntactic constraints brings out two points. First, the interpretation of an innovative verb is strongly constrained by its syntactic environment. This is hardly surprising. But second, these constraints do not work in a vacuum. To distinguish the interpretations of siren midday and siren the Porsche to a stop, one must know the difference between factory and police sirens, and how they are used. To interpret Ed’s teapot the policeman, one must know even more. So syntactic constraints must be considered along with all the other conditions placed on interpretations — conditions (a) — (e). No single constraint will suffice.