that are their Opposites: Noun to verb conversion in English
1. Background. In English, we find a rather large group of verbs that are formed from nouns with no change in form to the word. For instance, the noun perfume can be converted into the verb to perfume without undergoing visible change to the form of the word. To oil, to paint, to powder, to sugar are other examples of verbs formed from nouns.  This morphological process is sometimes called conversion (of one grammatical category to another), sometimes called functional shift (shift in grammatical function), and the verbs are often called denominal verbs to indicate their nominal roots.
verbs in English have been categorized into different semantic groups
(Clark and Clark 1979). For instance, some carry the meaning ‘to
go or move by vehicle’ as in to truck the tomatoes; others
mean ‘to make X into’, as in to powder the aspirin. Clark
and Clark (1979) provide an extensive corpus totaling approximately
1300 of such verbs in English. In this paper, I am concerned with
only locative denominal verbs, those verbs which express a locational relation of one noun to another. To oil
is one example of a locative denominal verb since it expresses the
movement and consequently locational relation of oil to (on) another object (hinges,
pan, and so forth). From Clark and
Denominal locative verbs regroup themselves into subtypes as a result of underlying syntactic differences (Marchand 1969; Clark and Clark 1979). If we look at the surface forms of the denominal locative verbs listed in Group A and Group B, we do not find much difference between the two sub-classes of locatives.
But if we look at a simple paraphrase of each of these verbs that expresses the relation of the parent noun to a more primitive verb put (a verb which captures the underlying meaning of all the denominal verbs in these two groups), we can observe some interesting differences between these two groups of locative denominals.  Group A verbs incorporate the underlying direct object into the denominal verb while Group B verbs incorporate the underlying noun which is in an oblique relation to the verb (it is the object of the preposition):
Here I give only a representative sample of verbs that belong to each subtype with the understanding that other denominal locative verbs that contain a put reading pattern themselves in such a fashion.
If we study the conceptual epistemology of these verbs as verbs of motion in general, we notice another characteristic difference about the relation of the verb in each group to its accompanying noun arguments. In Group A verbs, it is the object that moves (I will refer to this as the theme) that becomes the denominal verb: in to wax the car, for instance, an active agent causes the wax (the theme) to go from a particular unspecified point (the container?) to (on) the car, the specified stationary object location (I will refer to this as the place). In Group B verbs, on the other hand, we notice that it is the underlying stationary place object that becomes the denominal verb: in to bottle the wine, an active agent causes the wine (the theme) to go from an unspecified point (the vat?) to a specified stationary place location, the bottle.
In an article I wrote several years ago (Buck 1993), I explored the question of whether there still remained a semantic distinction between these subtypes that accompanied the underlying syntactic difference. I argued that the differences in the surface incorporation of underlying arguments in the subtypes are indeed systematic and can be explained on semantic grounds. I explained this by looking at the notion of affectedness of these alternations.
In group B verbs, if we ask the question of what has happened to the theme object after the theme arrives at the place location, the answer is that the theme has simply moved to another location. In to bottle the wine, for instance, the theme (wine) simply moves to the place location with respect to another previous location. And if we ask the question of what has happened to the place (bottle) as a result of the movement of the theme, we can say that it does not appear to be changed or affected in any way, its inherent properties of bottleness remain intact, independent of the theme.
In Group A verbs, however, the Place object is not quite the same after the theme has moved in relation to it. In to wax the car, upon arriving at the place location, the theme becomes a descriptive property of the place object and actually becomes part of the place rather than just simply establishing a locative relation to it. We can even comment on what has happened to the place object (i.e., after X waxes the car, the car is now shiny; after X greases the pan, the pan is now covered with grease; after X spices the food, the food is now spicier, and so forth). In Group A verbs, the theme and the place become one conceptual unit after movement. Group B verbs rather simply express a locational relation. So denominal verbs do hold constraints on their relation with the noun that surfaces as their direct object. We see a precise relation established between the theme and the place objects in each alternation.
So far I have presented denominal locative verbs that mean ‘to put one thing in or on to another’. These put verbs (Groups A and B) are called ornative verbs since they carry the meaning of adding one thing in or on to another (Marchand 1969:369). But locative denominal verbs also include a class of privative verbs (verbs that mean ‘to take away or remove one thing from another’). A curious regularity among locative denominals is that the privative verbs pattern syntactically in a similar way to ornative denominals and can be explained in terms of affectedness in a similar fashion.
To pit the prune can be paraphrased to to remove the pit from the prune while to mine the coal can be paraphrased as to remove the coal from the mine. Once again, we see that in Group A remove Verbs, the underlying direct object becomes the denominal verb, whereas in Group B, the underlying object of the preposition becomes the denominal verb. In addition, consistent with ornative verbs, the underlying theme is the denominal verb in Group A [an active agent causes the pit (theme) to move from the prune (place) to an unspecified point], while the underlying place is the denominal in Group B verbs [an active agent causes coal (theme) to move from the mine (place) to an unspecified point]. Group A verbs receive an affectedness reading: removing the pit from the prune changes and affects the prune (it is no longer whole or complete). Group B verbs, on the other hand, receive a location reading only with no change to the inherent properties of the place object: to remove coal from a mine does not result in any substantial change to the mine; the mine still remains intact even after coal is extracted from it.
2. Words that are their opposites. A curious phenomenon that occurs in English as a result of the fact that denominal locatives as a group are capable of producing either a put or remove reading is that we find certain words in the language that carry two potentially opposite meanings. The following list of paired phrases demonstrates that the denominal Group A verbs dust, feather, seed, string, trim, milk, tail, bark, bone, and cork, as examples, produce a put reading in (a) but a remove reading in (b).
(1a) to dust the cake pans (or crops) = to put dust on [i.e., flour dust; insecticide dust]
(1b) to dust the shelves = to remove dust from the shelves
(2a) to feather an arrow = to put feathers on the arrow
(2b) to feather a goose = to remove the feathers from the goose
(3a) to seed the lawn = to put grass seed on the lawn
(3b) to seed the grapes = to remove the seeds from the grapes
(4a) to string the guitar = to put strings on the guitar
(4b) to string the beans = to remove the strings from the beans
(5a) to trim the tree = to put trim decorations on the xmas tree
(5b) to trim the tree = to remove branches from the tree
(6a) to milk the tea = to put milk in the tea
(6b) to milk the cow = to remove milk from the cow
(7b) to tail the mouse = to remove the tail from the mouse 
(8a) to bark the leather = to put bark (an infusion of) on leather
(8b) to bark the tree = to remove bark from the tree
(9a) to bone the manure = to put bones in manure
(9b) to bone the fish = to remove bones from the fish
(10a) to cork the bottle = to put a cork on the bottle
(10b) to cork the oak = to remove the cork (outer tissue of the stem) from the oak
What, then, are the semantic/syntactic constraints that are operating on our understanding of these verbs? In other words, how do we know, when we hear one of these ambiguous words, whether the verb means ‘to put X on’ or ‘to remove X from’?
One way to explore the meaning of these verbs is to examine the world knowledge that is assumed in our understanding of the denominal verb. If we systematically trace and describe the processes, events, and states associated with our knowledge of concrete objects and look at them in relation to what we know about space and time and basic physical laws, we arrive at a better understanding of the relation between the verb and its arguments (Clark and Clark 1979:769). It is clear that the semantics of these verbs is relational rather than inherent. But which part of the meaning is inherent to the verb? How much? Where is the boundary? What world knowledge restrictions operate between the verbs in these cases and their direct objects? What restrictions do these verbs place on the set of possible objects they choose?
3. Characteristics of remove verbs. If we consider the following examples of Group A remove verbs, we find that typical of privative verbs in general is that, in the real world, we find the objects which are represented as the theme existing in some sort of natural affected state prior to the denominal action of the verb (Marchand 1973:637). Most are part-whole relations.
However, the more important point is that the objects which are themes cannot possibly be put where they are by an active agent since they occur naturally in this state. For instance, in the example to bone a fish, an active agent cannot cause the event of putting bones inside a fish. One cannot put bowels inside a squirrel; or burls on a cloth; or a core in an apple; or a fin on a fish. So the action of Group A privative verbs is constrained by the nature of the relation between the selected themes and places. The action that an active agent can undertake can go only one way. An active agent can only remove the theme from the place object; the reverse put action can never be performed. 
4. Words that are, but aren’t, their opposites. If we go back to a few of the paired examples of Group A two-meaning verbs listed in Section 2, we notice that each verb does not really potentially carry, after all, two directly opposite meanings. In the first example of to dust, the underlying place nouns in the paired examples refer to different objects in the real world, and so do the underlying nouns acting as themes. Dust in the put reading refers to flour dust or insecticide dust; dust in the remove reading refers to the dust that simply occurs naturally on objects in our environment. It is important to note that the denominal privative verb to dust never means ‘to remove flour dust from the cake pans’ or ‘to remove insecticide dust from the crops.’
The same holds true for examples (3), (4), (7), (8), and (10). To seed in (3b) means ‘to remove the seeds which naturally occur in grapes’; to string in (4b) means ‘to remove the strings which naturally occur on beans.’ To seed as a remove verb never means ‘to remove grass seed from the lawn’; to string never means ‘to remove strings from a guitar.’
In examples (2) and (9), the underlying nouns acting as themes refer to the same objects in the real world, but the place referents are different. In to feather, as representative, the feather in (2b) exists in a natural state on the goose; in (2a), the feather does not share a natural state relation with the arrow. The privative denominal verb, to feather, then, never means ‘to remove the feather from the arrow’.
The verb to trim in example (5) is a bit more ambiguous because the underlying place nouns in (a) and (b) refer to potentially the same object in the real world. Even though (a) could refer to an artificial tree, it may also refer to a “real” tree. However, again, the phrases are not truly ambiguous since the underlying theme nouns refer to different objects: in (a), trim refers to decorations, ornaments, and so forth; in (b), trim can only refer to tree branches that occur in a natural state on a “real” tree. Once again, the privative denominal to trim may not mean ‘to remove the Xmas decorations from the Xmas tree.’
The point here is that the reverse action of a privative verb is not the action characterized by an ornative verb. A Group A remove verb does not mean the reverse action of a Group A put verb simply because put verbs select objects that active agents can put on or in the intended place objects.
Dusting a cake with sugar, for instance, is something an active agent can do. Sugar on top of a cake does not occur naturally; the natural state of a cake does not include powdered sugar on it. Therefore, dusting a cake is a well-formed ornative verb but is disallowed as a privative verb; the constraint of natural state relation between theme and place nouns disallows cake from being selected as the place object of the privative verb.
To cork the bottle is an interesting example to look at. To cork the bottle meaning ‘to put a cork in the bottle’ is a well-formed ornative verb. An active agent has the ability or power to cause the cork to be moved from Point A to (on) Point B (the bottle). However, the reverse action would be simply to remove the cork from the bottle—move the cork from Point B back to Point A. But to cork the bottle is an impossible privative verb. Since the cork top does not occur in a natural state on the bottle (a state which may not involve an active agent), to cork the bottle can get only an ornative reading.
In the example to trim the tree, when the theme refers to decorations, we realize that an active agent can move decorating trim onto a tree; the decorations do not occur in a natural state on the tree. This allows for a well-formed ornative reading. Trimming the tree may not receive a privative reading (‘to remove the decorations from the tree’) as long as the referent of the theme remains decorating trim. But trim can also refer to tree branches. Since tree branches occur in a natural state on trees (an active agent being unable ever to put them there in the first place), trimming the tree may receive a privative meaning.
We are constrained then not to interpret dust the cake as ‘remove dust from the cake’ or feather the arrow as ‘remove the feather from the arrow’. The constraint on privative verbs is that we can only remove things from objects that have an initial natural prestate that does not involve an active agent. For a charted illustration of denominals and their opposites, please see the Appendix.
5. Reversing Group B remove verbs. As mentioned in Section 1, Group B privative verbs 1) incorporate the underlying object of the preposition (the place noun) as denominal verb; and 2) carry a simple location reading rather than an affectedness reading to the place object. Consistent with the additional constraint of natural state relation between theme and place nouns, privative Group B verbs are consequently limited in number in English. Only two denominal remove verbs that I have found denote removal of an object from a particular place where the place is a natural place location and where removal of that object from the place location does not affect the place location. They are to mine and to quarry.
We note, then, that we have no privative denominal verbs which mean ‘removal of an object from a place location that is not found in nature.’ The verb *to dresser, for instance, meaning ‘to remove the clothes from the dresser’ would be an impossible privative verb in English since clothes and dresser do not share a natural state relation. Active agents are responsible for putting clothes in dressers; the clothes do not just exist there naturally.
The reverse action of Group B privative verbs is not possible since an active agent simply cannot put coal in a mine or marble in a quarry. Those objects exist in their natural states in these natural place locations.
6. Reversing Group B put verbs. This leads us to the point that the direct opposite of a Group B ornative verb, such as bag, cage, coop, and box as examples, cannot be a denominal verb. The verb in *to bag the groceries meaning ‘*to remove the groceries from the bag’ is an impossible denominal verb, not because the ornative verb to bag already exists but because bag does not meet the natural state requirements of privative verbs. In our lexicon, the vehicle we use instead to capture the opposite (reverse) action of put denominal verbs is the prefix un-, as we see from the following examples. 
The prefix un-, however, does not do a complete job of reversing all Group B ornative denominals. Even though to unskewer the meat, to unbottle the wine, and to unspool the threat are all possible verbs, they do not seem to be used. Likewise, the following are a few examples of verbs that appear to be accidental gaps in the language. Some are logically strange, but a few could be possible verbs but do not seem to be used.
7. Reversing Group A put verbs. Group A ornative verbs capture different degrees of affectedness on the place object after movement of the theme has occurred. Some verbs such as in to saddle the horse, to newspaper the shelf, and to mast the ship result in the theme and place forming one conceptual affected unit, yet the theme and the place do not converge as they do in such verbs as to sugar the tea, to wax the car, to oil the pan, etc. The place remains intact, but affected, in these Group A verbs of least affect. The reverse action of these Group A ornative verbs is captured by the prefix un- (yet another function of the prefix). The following verbs represent a few examples of Group A ornatives of least affect that have reverse actions captures by un-.
Some Group A put verbs simply cannot logically reverse. A few examples are sugar the tea, butter the bread, dust the cake, salt the fish (meaning salt with a salt shaker). An active agent simply cannot remove the sugar from the tea once it has been put in the tea; every bit of butter cannot be removed from the bread; every flake of powdered sugar cannot be removed from the cake once it has been sprinkled on; every morsel of salt cannot be removed from the fish. The place object becomes so affected by the theme noun after movement that the theme and the place objects essentially become one unit. Consequently, no verb that captures such a reverse process is in the lexicon.
Other Group A put verbs also result in the same complete affective state, but they can reverse simply because some very extensive process has been devised in the real world to remove the theme object from the place. We note that the following Group A ornative denominal verbs are actions that are carried out by an active agent, and the action in each case results in a complete affective state of the place object: to grease the wool; to gum the silk; to lead or silver or tin the object. The reverse activity is, of course, not captured by a denominal verb since an active agent can put grease, gum, lead, silver, or tin on the place object (a condition not allowed by privative verbs). In the lexicon, the reverse activity of these verbs is captured by the prefix de-: to degrease the wool, to degum the silk,  to delead or desilver or detin the object. I believe one can desalt fish, meaning ‘to remove salt that was used to cure or preserve the fish’. So one of the functions of de- is to capture reversal of affective states of natural place objects that have been affected by an active agent.
Interesting about the prefix de- is that it has an additional function: it overlaps with the function of privative denominal verbs in that it also captures the meaning of ‘removal of an object from a preexisting natural state (that may not be put there by an active agent)’: dehorn the cattle is an example, as is defrost the refrigerator. De-ice the plane, delint the clothes, desalt the water (the urine, the solution), defat the antigen, deforest the hill, and degrain the leather are other examples. 
result of this overlapping function with privative verbs is that
we find in the lexicon redundant pairs such as the following: 
8. Conclusion. A study of denominal verbs offers a microcosmic view into our knowledge of the semantics of verbs in general. My interest is to attempt to specify the type of unconscious semantic knowledge that native speakers have that allows them to make distinctions between selected types of denominal verbs. But the interests are driven by larger and more important questions: How do we organize our mental lexicon of verbs, and what connection does this reveal about language and the way we perceive events, states, processes, and objects in our everyday world? As Jackendoff (1983) states, the study of meaning in language can, after all, reveal something to us about “what we perceive and what we do” (p. 3).
Most denominal verbs, as we have seen, are characterized by their concrete nature: they involve very common objects and events in our everyday lives (sugar, milk, paint, powder, fleas, weeds, for instance). Most curious, though, about these verbs as a group is that a very large number of common nouns cannot be converted into denominal locative verbs. We do not say, for instance, *to bowl the milk, *to apartment the dog, *to pineapple the cake, or *to thyme the soup. Verbs like *to handkerchief, *to bra, *to teapot, *to tablecloth, *to pillow, *to sofa, *to window are not denominal verbs. What constraints are operating, then, on our understanding of locative denominals as a group that allows us to know which nouns are easily converted and others disallowed from the group?
It is curious that many of the common nouns that convert into denominal verbs have etymological origins that predate Anglo-Saxon. Most of the privative verbs, in fact, noticeably deal with agricultural activities (some of which are obscure to us, like barking leather, burring wool, and burling cloth). It is possible that some sort of historical constraint is operating on these verbs; perhaps the answer is to be found in a study of Indo-European roots of these locative verbs as a group.
 I will always be indebted to Beth Levin for getting me “hooked” on denominal verbs.
The title of this paper, “Words that are their opposites,” was used as the indexing title of an Internet on-line discussion which took place on LINGUIST in January and February, 1995. I am using the title here because it nicely captures the essence of these verbs, but I do not know who originally coined the phrase; I would like to give full credit here to the person who did. The ideas in this paper do not relate to that discussion, which focused mainly on semantic shifting rather than on conversion, although a handful of examples of denominal verbs were mentioned in that discussion.
 Consistent with Clark and Clark (1979), included in the group of denominals are noun to verb conversions that involve final voicing shifts (p. 768). The noun shelf, for instance, undergoes a slight change in form as a verb, to shelve, but the verb is still included as a denominal since the change is due simply to voicing.
 As in my former study (Buck 1993), I continue my interest in the literal, concrete meanings of these verbs. Clark and Clark’s corpus includes innovative denominals, those that show metaphorical and sometimes abstract uses of the verb. Some of the verbs in my corpus carry extended meanings, too, but no attempt has been made in my analysis to capture the meaning of each individual denominal verb in its entirety. I am interested, rather, in the basic meaning that underlies all the verbs in this class, with the understanding that extended meanings are derived from these basic underlying meanings.
 See Clark and Clark (1979:769) and Duszak (1980:51) for a discussion of the use of the paraphrase method in capturing the underlying semantics of denominal verbs.
 One interesting exception I have found is to fish the stream, for 1) one can certainly stock a stream with fish; and 2) removing fish from the stream doesn’t affect the stream in the same way as the other privative verbs in this group.
 I have found a few exceptions to the functions of de- as outlined in this paper. In the following examples, de- appears to be overlapping with un- in capturing the reverse action of Group B ornative verbs: to dethrone the prince; to detomb the corpse; to debar the lawyer.
 To ungum the silk
is also an existing verb in the
 For more on the functions of un- and de-, see Colen (1980/81) and Funk (1990).
 Many of these examples are from Ross (1976).
Buck, R. A. 1993. “Affectedness and other semantic properties of English denominal locative verbs.” American Speech 68.2:139-60.
Clark, Eve V. and Herbert H. Clark. 1979. “When nouns surface as verbs.” Language 55.4:767-811.
Colen, Alexandra. 1980/81. “On the distribution of un-, de- and dis- in English verbs expressing reversativity.” Studia Germanica Gandensia 21:127-52.
Funk, Wolf-Peter. 1990. “On the Semantic and Morphological
Status of Reversative Verbs in English
and German.” Further Insights Into Contrastive Analysis. Ed.
Duszak, Anna. 1980. “A Semantic Description of English verbs with Semantically and Formally Related Nominal Counterparts.” Kwartalnik Neofilologiczny 27:49-71.
Jackendoff, Ray. 1983. Semantics and Cognition.
Marchand, Hans. 1969. The categories and types of present-day
English word-formation: A synchronic-diachronic approach. 2nd
1973. “Reversative, ablative, and privative verbs in English, French,
and German.” Issues in linguistics.
Eds. Braj Kachru.
Ross, Alan S. C. 1976. “Meaningless ‘de-‘ in English.” Notes and queries 23:310-11.
DENOMINALS AND THEIR OPPOSITES