(Lecture for the Berkeley Summer Research Seminar, George Lakoff, Director, June 1996.)

Price Caldwell

Hino-shi Ishida 1-12-6

Twin Column Tama B-201

Tokyo 191-0022, Japan

The Molecular Sememe: A Model of Ordinary Language

(And mustn't it also be a model of human cognition?)

.;

I. Introduction:

Some years ago, I attended graduation exercises at a small American university. During the course of the ceremonies, the university Registrar said, "This is undoubtedly the largest class we have ever graduated." I began to wonder what the word "undoubtedly" meant.

If he had said, simply, "This is the largest class we have ever graduated," I would have taken it as a statement of documentary fact, coming from the authority charged with assembling such information. Since he didn't say that, I suspected that he didn't know for sure--had forgotten to look at the actual numbers--and was guessing. But since he also did not say, "This is probably (or most likely) the largest class we have ever graduated," I concluded that he wanted to pretend he was not guessing. To put it briefly, his use of the word "undoubtedly" created in me a good deal of doubt as to whether he knew this was indeed the largest class the university had ever graduated.

But while his statement put itself in doubt, it also suggested a rich complex of meaning. There was a range of nuance which went even beyond his own communicative intent. The meaning that I came to understand, whether correct or not, was antithetical to the dictionary definition of the word--

un-doubt-ed (un dou'tid), adj. not doubted or disputed;

accepted as beyond doubt. [ME] --un-doubt'ed-ly, adv.

--Random House

--and was a meaning which involved, in a negative way, the meanings of the other terms which could have been used in its place. Whether consciously or not, the Registrar had considered and rejected those words, and therefore the meaning of the word he did choose meant what it meant by contrast to the other terms in that little momentary "molecule" of possibility. It occurred to me that this tiny and shortlived but dynamic set of choices, within which a specific and highly concrete and subtly nuanced meaning had been born, might itself be the "sememe," the fundamental unit of meaning, and the ordering principle on which a semantic theory could be based.

Ø

probably

This is undoubtedly the largest class we have ever graduated.

most likely

perhaps

etc

Now, as a fiction-writer and literary critic, I've always wanted a theory of meaning that accounted for subtle, suggestive, complex or many-layered meanings such as those found in literature. Linguists' ideas of meaning are usually minimal-- at worst, truth-conditional meaning, or at best, conventional dictionary-standard ("literal") meanings. The idea of a molecule appealed to me, therefore, because it seemed to account for a subtle, many-layered meaning that was however controllable and discussable. It seemed to allow for, and even suggest an account of, the way implications are born and controlled. To see meaning as belonging to the molecule AS MARKED BY the chosen word, rather than to the word itself, seemed to put the problem of meaning in an entirely new light.[1]

I. 1. Intuitive attraction of the theory.

Further thinking led me to other examples. Why is it that the color "white" changes with every new context? Why is a white dog a different color from a white man or a white house or a white lie? How do we know when it is being used metaphorically? Could it be because the DISCOURSE the word presents itself in always selects its contrastive possibilities differently? One discourse might select a "white" man as opposed to an hispanic or oriental or black man --

{WHITE/ hispanic/ oriental/ black} man

while another discourse may select a simpler set of oppositions in which "white man" is as opposed to "brother"--

{ WHITE MAN / brother }.

And a {WHITE / brown / brick } house is probably of a different color from a {WHITE / brown / yellow / black } dog. And a {WHITE / dirty } lie is of a different "color" from any of the above.

It also made sense to me to think of the many meanings of "out" in similar fashion. A light can be {OUT / on }, and a baseball can be {OUT / in }, and the baserunner can be {OUT / safe }. An accident victim may be {out / conscious } and the office workers may be {out of / in } the office.

And the other day I found a product in the pharmacy that interested me: "Non-Aspirin." I wondered what was in the box. By categorical logic, it might be anything that was not aspirin: sharks' teeth? marinated goose liver? Ground cumin seed? But of course it was acetminophen--generic Tylenol. Though it was not Aspirin, it was in the same molecule as aspirin-- {aspirin / non-aspirin } and no less a pain-killer than aspirin.[2]

And then I remembered my mother, calling her daughters. She'd call any child by the name of any one of them. "Delia--Jane--Marg--Anne? Whichever one you are . . .!" It was clear to me that the realest thing in her mind was a category of daughters--and that it could be named by calling any one of the names. Naming exactly which one she meant seemed to take an extra degree of mental effort. Obviously this category of "daughters" was not a formal category, but a molecule: a synthetic structure whose coherence was very real though never really named except by the name of any one of its members. It seemed obvious to me that the human mind must be far more attuned to synthetic structures like these than to the analytic categories of science and formal logic. Likely, I thought, analytic categories are special cases of molecules, derived from them rather than fundamental to them.

I. 2 History

The basic notion underlying this "molecule" is, of course, not at all new. It is a revision of the notion of the "contrastive set" or the "paradigmatic series" familiar to early Structuralists in the Prague School tradition. The contrastive set was a local, context-bound expression of the famous Arbitrariness Principle: every linguistic unit found its value not in any particular acoustic properties but in its contrast to every other unit. Unfortunately, linguists were never able to make contrastive sets fit in with any of the categories of syntax, suggest any consistent structures of their own, or even to yield up their common features. In frustration, the effort was abandoned. Early on, deciding that speech was just too heterogeneous to deal with, Saussure had appealed to the idea of Language as an idealized set of relationships instead. By that means he hoped to "introduce a natural order into a mess that lends itself to no other classification (Saussure, Course in General Linguistics, 9).

The rest is history. Linguistic science chose to follow the path of general science, seeking the properties of a single structure called Language as if it were an object in physical nature like a table of billiard balls. Roman Jakobson set out to make lingistic science as respectable as Physics by tearing down the arbitrariness principle. By 1955 he was convincingly arguing that phonological structures around the world could be accounted for by some twelve inherent features and a few prosodic features. Through feature analysis, Jakobson reduced the differential character of the phoneme to a bundle of binary oppositions. The phoneme became a mathematically simple set of plus or minus markedness values, and Saussure's arbitrariness doctrine disappeared in favor of a computational taxonomy.

Thus the concept of the differential gave way to a theory of distinctive features as an information-encoding structure which could be merged with purely mathematical notions of communication theory. The resulting notion, that the system of language could be described by logical and automatic rules, encouraged Chomsky's proposition that Saussure's idealized social object was in fact a set of competency rules unconsciously internalized by the native speaker (Chomsky 1965, 28-29). Later, Jakobson finished destroying the principle of the differential at the morphological level, again through markedness theory, in which pairs of lexical items were seen to stand in motivated, asymmetrical, binary oppositions to each other. Language began to be seen not as a structure of arbitrary relations, but as a "code" in which binary oppositions at the bottom level generated meanings through a series of generative rules. Since computers are very good at manipulating binary codes, general optimism could easily foresee the day when the generative rules would be exhaustively discovered for semantics as well as syntax, the code could be completely machine-managed, and Artificial Intelligence would help us all get our work done.

Still, however, there was no successful semantic theory, and no connection between language and general cognitive functions. As a result, context-oriented theories of meaning were revisited by second-generation cognitive linguists who realized that meaning must involve more than truth-conditional semantics, that it must take into account communicative objectives (performance issues), and that it cannot be understood apart from context. The effort to formalize contextual constraints led to the invention of quite a number of cognitive structures, such as frames, scenes, cognitive domains, folk models, scripts, schemas, and idealized cognitive models.

Since this is precisely the point at which my model conflicts with the model proposed by Lakoff and Johnson, let me try to put the conflict carefully. First, the cognitive terrain is well worth studying on its own account, because it can provide an accurate and consistent mapping of structures of thought underlying a number of intellectual positions in science, politics, sociology and popular culture. The study of metaphor, for instance, proves that interior experiences, whether emotional or intellectual, must always be mapped onto orientations borrowed from the exterior, spacially-extended world in order to find any articulation at all.

The question I have is whether it really pre-exists and motivates ordinary language. I wonder if it isn't more likely the case that the sense of a cognitive unconscious derives from discourse.

Beyond that, many have tried to map the many domains of the cognitive terrain because they assumed that it pre-existed language, and that these domains could provide a context within which linguistic expressions could find meanings richer than truth-conditional or "literal" meanings.

First, I recognize that these tactics are motivated by assumptions which are taken to be fundamental to the scientific enterprise. Among these are (1) that the object of study, language, is taken to be a single entity or a single system of entities, in the sense that any true proposition about any part of it must apply equally well to any other part of it (the law of noncontradiction). And (2) that it must be analyzable into formal constituents so that empirical experiments such as minimal pair analyses can be conducted. On the other hand, the assumption that there is a level of (unconscious) cognition which pre-exists language may be only a carry-over from the Structuralist concept of La Langue, in that though it is NOT language, it is like language in that it is a single structure of relationships to be studied as a single synchronous order. But if that were true, then in order to formalize the possible context-constrained meanings in a cognitive field, one would have to provide an exhaustive and encyclopedic listing of all the domains, frames, or scripts in it.

That seems to me an impossible task, and I wonder whether these assumptions are really necessary. It seems to me they make it impossible to study language as anything other than a single synchronous structure. If meaning in language is owing to the principle of the differential within specific discourses, however, then meaning must be essentially pluralistic, and any analysis requiring formal categorization would destroy it as a principle.

It therefore seemed to me that the "molecule," rather than the formal category, might serve as a description of the order Saussure was looking for. Because its contents are selected by all the hetergeneous factors which go into speech, it must also be an index to all that heterogeneity. I resolved, therefore, to keep the idea of the differential as expressed in the "molecule," and to try to criticize every principle, especially methodological ones, which conflicted with that idea. It required that I could no longer treat the language as a single system at all. The molecular sememe is a unit of meaning in discourse. It seemed necessary to try to develop what Levi-Strauss called a "science of the concrete."

II. Principles of a science of the concrete

--Performance precedes competence (La Parole precedes La Langue). Discourse precedes grammar. Pragmatics is not a second-class realm.

--Analysis must be of actual discourse, not "sample sentences."

--No formal, analytic structures. No formal categories. Only molecules.-- (Categories are a special case of molecules.)

--No atomistic primitives (features, figurae, phones, etc). We abandon the constituency theory, or the principle of analysis by reducing everything to constituents, on principle.

--No automatic, computerable, generative rules. I do believe in generating text, but through human response and choice, not by program.

--Syntax is not primary or fundamental. Discourse salience and molecule selection are fundamental; syntax is derived from them.

--Meaning is discourse-dependent and pluralistic. All rules and structures are local; there are no universals.

III. The Model:

1. The basic unit of meaning is the Molecule. { }

a. The molecule consists of whatever choices are available, given the context, at any decision-point in the generation of a discourse. Almost always, this decision-point occurs at the introduction of new information into the discourse.

b. The meaning belongs to the molecule as marked by the word chosen to name it.

c. The contents of the molecule may be synthesized by the mind in many different ways. These various dynamics make the familiar rhetorical structures possible (See V.b. below).

2. The basic unit of discourse is the Molecule Selection-and-Execution Structure (MSES). --> --> --> { }

1. The contents of the molecule are selected by all the elements of discourse which precede it. These include non-verbal factors such as scene, speaker, spoken-to, circumstances, motivations. Thus the molecule is an order of possibility created by context.

2. This context is expressed by the mind as a synthesis of words in an ever-increasingly marked order in the syntagmatic plane. To put this another way, the words which precede the molecular term, though less marked than it is, conspire in their grammatical relations to constrain the possibilities of the molecule itself.

3. The words which comprise the molecule-selection structure of the discourse may themselves have marked molecules at earlier stages in the discourse. Fully marked then, they now move to less-marked status, and operate to limit the possibilities of the next molecule.

3. Discourse is a hierarchical order of MSES's.

--> { }

|

--> -o- { }

| |

--> --> -o- { }

| | |

--> --> --> -o- { }

Simple discourse as example:

Call me { Boris }.

famous

My name is not { important }

relevant . . .

best dream

What's important is that I am going to become your {worst nightmare.}

-- Items which were most salient in previous sentences may have receded to immediate background or deep background for the current sentence. While their meanings are less highly marked now (indeed they will have shrunk to their most conventional meanings), they still continue to affect the selection of the contents of the next molecule.

IV. This research has three important parts: Since the Molecular Sememe gives rise to both lexemes and taxemes, I must treat semantics and syntax separately.

1. How the molecule creates meaning. A theory of Semantics.

2. How the molecule can become a grammatical category. A theory of Syntax.

3. How the MSS selects the molecule. A markedness theory of discourse.

IV. 1. The molecule as Semantic Structure:

Supposed you found a discourse in which the following sentence occurred.

He opened his eyes and discovered to his great surprise a girl with feet.

It should be obvious that this is not a successful piece of discourse, that something else is needed to give it a point. What kind of feet? Clearly, the most highly marked term in this sentence is the word "feet," but the thing that should mark it is missing.

He . . . discovered . . . a girl with { } feet.

With only one sentence, one has a lot of freedom in picking the adjectives that might qualify "feet." There is little sense of constraint.

However, when each choice is selected by a larger and denser isomorphism of meaning and implication, the choices narrow considerably. To illustrate the point, here is a larger sample of the context of these molecules: in this story a man named Gordon took a can of beer and got into a warm bath on a Sunday morning in February, in the dead of winter, and promptly fell asleep.

In the warm water his body began to clot like an egg and he napped until someone touched him lightly on the knee. He opened his eyes and discovered to his great surprise a girl with tanned feet. She was sitting on the ledge just above the bathtub. Her feet were propped one on top of the other, toes curled on the spigot. Her hands were clasped around her knees. He clenched his eyes in disbelief and took another swallow of his beer. When he opened his eyes she was still there. She wore a light cotton summer dress, printed with flowers.

"Hi," she said brightly. (Baby Bluefin pp. 2, 3)

The word "tanned" is necessary: the anomaly is that it is wintertime, yet the fantasy woman wears a light summer dress as if she is fresh from the beach. We might guess again about the contents of the molecule:

pale

white

. . . a girl with tanned feet

stockinged

Since the molecule is selected by context, this means that the selected counter is not just a term, but a term-in-context. Thus, its meaning can, in the hands of a creative maker of texts, be fully nuanced, rich, and complex in implication and application. In this case, TANNED = Foot-descriptor: not weak, not thin, not stockinged but TANNED.

This example should illustrate several principles:

--The molecule is selected for a specific position in a specific discourse.

--The molecule, while it often consists of conventional pairs of opposite terms such as {yes / no }, {cold / hot} etc, may consist of possibilities which have never been juxtaposed before. In the hands of good poets, such molecules create new meanings for familiar words.

--The meaning belongs to the molecule as marked by the chosen term, not to the term itself--though it will get credit for that meaning.

--Since meaning does not belong to the word itself, the arbitrariness principle still holds.

--The molecule consists of contents, not just signs. It is the arena in which words are forced to match up somehow with the world. (Supposed you are a young bachelor and you've been often been invited to supper at the house of a married couple-- You arrive at the door this time and your host says, "Honey--guess what he brought us THIS time!" Clearly the question points to a molecule which consists of hostess gifts. And the meaning of this gift depends on the molecule of all past gifts AS MARKED BY the present gift.)

--The molecule is selected without the speaker or hearer necessarily being aware of it. Its creation reveals the action of the unconscious in the creation of meaning. (This is a very important point. The question of the role of the unconscious in discourse is a very difficult one. But clearly both speakers and hearers must unconsciously construct the molecule from which one item is chosen, and his sense of the meaning is dependent on that unconscious content.)

--As the hearer may unconsciously construct a different molecule than the one the speaker intended, misunderstanding frequently occurs. (Another very important point. One person may use the word democratic and mean as opposed to aristocratic; it may be HEARD, however, as Democratic as opposed to Republican. My good friend Peter Shillingsburg has a theory of textual editing in which he borrows this idea to talk about the Molecule of Origination and the Molecule of Reception, to show how people misunderstand each other by unconsciously constituting these molecules in different ways.)

--The discourse as a whole may consist not of words but of events, as long as it can be constructed as a molecule selection-and-execution structure (MSES). Thus we can interpret non-verbal events, as well as music, dance, and art, to have meaning. In the same way, Ludwig can interpret human behavior and know what his master is doing. This suggests a theory of animal communication as well as a way of describing the discourse of art or music.

Vb. The various internal dynamics of molecules define the universal rhetorical structures. This is further evidence that the molecule is fundamental to language. All languages have the structures of assertion, interrogation, naming, generalizing, itemizing, etc.

1. When you have read the question, mark the letter corresponding to

the answer as given in your exam booklet: A, B, C, or D. (Interrogation and assertion).

A fully explicit, symmetrical molecule such as the molecule [A/B/C/D] interrogated here is indistinguishable from (and probably the origin of) a logical set. Purely logical counters are typically members of explicit, symmetrical molecules. The molecule [question/answer] is also fully explicit and symmetrical. This kind of molecule characterizes the language of technicality.

2. Q: Coffee, tea, or milk?

A: Coffee, please.

The molecule [coffee/tea/milk] is an itemized and fully explicit molecule. If it is, it suggests the logic that once obtained in the universe of airline food, within which coffee, tea, and milk were the only alternatives. As should be apparent, the primary rhetoric of interrogation is to itemize the contents of the molecule. At the same time, the primary rhetoric of assertion is to name an item as selected in place of others. Hence, assertion is implicitly a denial of the unselected counters.

3. Q: Did you say 423-2345?

A: No, I said 423-2355.

This molecule, which will be indicated by a tone change emphasizing the new "5" in the answer, is [--- --4- / --- --5-]. That is, it is implicit that the molecule is an order of telephone numbers. Thus, a molecule can overlap with or include a set of numbers. Of course, the "dynamics" here are very logical, and do not possess the nuance that an organization of non-abstract terms possesses.

4. Q. When did your plane arrive?

A. It didn't. I had to take a bus from D.C.

Here the molecule proposed is a WHEN molecule including all the possible times a plane might arive {TIME}. But it is rejected in favor of renaming the molecular term as {plane/bus}

5. I won't take 'no' for an answer. (Negation)

[I won't take 'no'/(I will take 'yes')]: this molecule should illustrate that the typical molecular strategy of negation is to deselect the explicit term and therefore select the opposing term.

6. No, Mr. Smith is not the one I had in mind.

The molecule here is [Smith / whoever else IS held in mind] If the hearer can be relied on to know the possibilities of the molecule, then negating one term effectively names (without naming) the other(s). Gossips and news leakers often use this method of saying without saying.

7. Either fish or cut bait. (Analogy)

This kind of explicit alternative proposition is really a kind of explicit negation, implying that if you reject one alternative you are stuck with the other.

At first glance, [fish/cut bait] might seem, on an abstract level, reducible to a simple categorical set (YES/NO). But that analysis, in ignoring its "poetry", would be inaccurate. The molecule really invokes a very rich universe of values (which we might call the language of fishing) within which a far more particular logic operates. Generalized and paraphrased, it means something like "either engage the main task or support those who are so engaged." But I only know that because of the analogical relations that are possible between molecules of similar shape.

8. "T'is not so wide as a church door, nor so deep as a well,

but t'is enough, t'will serve." (Mercutio, in Romeo and Juliet)

Another metaphor. It is important to notice that it is generated by the same process by which any other naming can occur. That is, the "it" of the discourse, the seriousness of Mercutio's wound, is never named except by negation. As in many metaphors, the unconscious must supply the unsaid measures of the width and depth of the wound.

9. Why don't you put a voltmeter on the starter solenoid and see if

it's getting any juice? (Metaphor)

If we look at the "molecule-selection" parts of the sentence, we find so much redundancy that the molecular term, we strongly presuppose, will be "electricity." In short, any sort of token that could be possibly related to the idea of electricity could have been put in this slot. Its value, in turn, will be measured by its differential from the word "electricity." "Juice" is clearly more fun than "voltage" or "current" or "energy" would have been, and, because any term would be interpreted by analogy to electricity, there is no loss of precision. Even counters such as "tingle", "flow", "stuff", "oxygen" might be close enough to work.

10. He: Where have you been?

She: I just went out to get a breath of air.

He: You did, like hell.

She: What do you want me to say, Darling?

He: Where have you been?

She: Out to get a breath of air.

He. That's a new name for it. You are a bitch. (Naming)

(Dialogue from Hemingway's "The Short Happy Life of Francis Macomber")

An implication here is that any token can be made the name of anything if it is put in the position of a strongly expected term (here the reader has been strongly prepared to understand that the Female Speaker has been out having sex with the Other Man). As in the previous example, the semantic value of "out to get a breath of fresh air" is given in the differential between what those words ordinarily mean and what they mean now, which is, "in bed with Mr. Wilson." This differential adds up to the insouciance with which she (Margot Macomber) equates the one with the other, without any real effort to obscure the true referent. It is what provides the definition of the word "bitch" in the final act of naming.

To summarize:

Molecules operate with several different kinds of dynamics:

Naming one counter can negate the rest (assertion).

Negating one counter can implicitly choose the rest (negation)

Naming one of the counters can name the molecule (synechdoche)

Naming the molecule implies all its contents (generalization)

Itemizing the molecule specifies its contents (itemization).

A Question Word refers to the molecule and queries its contents (who, when, where, why, what).

These rhetorical possibilities are universal. I claim this as evidence that the molecular structure is fundamental.

V.c. Molecules as Syntactic Structures: Grammatical status belongs to the molecule as a whole, not to its counters.

Another way to say this might be simply to observe that all members of a molecule have the same grammatical value. But this value is conferred by membership in the molecule, not already present in the terms themselves. The counters in the molecule in IV. 1. above, for instance, do not have any formal grammatical status in common.

thin

white

1.b.a . . . a girl with tanned feet

stockinged

"Stockinged" began as the noun "stocking," and then was marked with a participle-making affix (-ed). to create a past participle/ adjectival form for which there probably was never a verb. "Tanned" derives from an adjective ("tan") by way of a verb form. "White" and "thin" are uninflected forms usually described as "adjectives," but the point is that their eligibility for membership in this molecule is not dependent on their original forms. Rather, their grammatical status is conferred upon them by their membership in this molecule. One might say that whatever their stem forms, they have been marked or tagged with whatever affixes render them able to participate in this molecule. Meanwhile the molecule itself, which has no name and may well exist for this sentence alone, can be described in any way we like--as a "foot-descriptor molecule," for instance. In any case, the molecule itself is the unit which can properly be said to possess grammatical status.[3]

--The molecule is selected for a specific place in a specific discourse.

--All the contents of the molecule have the same grammatical status, whatever that is. This really means that the grammatical status belongs to the molecule as a whole, not the individual term.

--The grammatical status may not be namable, but as a marked term it may have a very specific grammatical status.

--The molecule is not a formal category but a synthetic structure. A formal category is a special case of symmetrical, explicit molecule when the typical molecule is asymmetrical and inexplicit.

--As the chosen terms move into their unmarked positions, they may lose grammatical specificity, and thus be more easily generalized as similar to other terms and thus put into conventional "categories."

One argument is that grammatical categories as presently conceived aren't very solid. What, for instance, do these direct objects have in common??

Jennie told her parents she was going to Anne's house.

She said, "I won't be late."

Jennie shared her thoughts with Anne frequently.

She hoped Anne would understand.

But Anne often pled ignorance of Jennie's meaning.

VI Principles of Molecule-Selection Structures:

--Every discourse has its unique molecule-selection structure, which separates it from all others.

--Every Discourse proceeds from least-marked to most-marked terms, more or less linearly. When this order is violated, the result seems, though not ungrammatical, somehow vaguely un-English.

--This order is a salience order. Whenever more than two or three signs have to be communicated, it stands to reason that one must communicate not only the signs themselves, but the order in which they are to be taken as relevant. This order expresses the intentional nature of the discourse as motivated speech.

--This order is communicated by many things: rhetorical structure (narrative and metaphorical committments, etc.) and by ad-hoc markedness promotions or demotions. Indeed, syntax is a means of indicating such marking this rhetorical structure, but is not sufficient by itself.

VII. Recapitulation & conclusion.

1. The model implies an account of language learning in children.

2. The model implies an account of language change over time.

3. The model implies an account of communication in animals.

4. The model implies a semantic theory which may be able to account for imports, implications, tones, styles, and voices as essential parts of meaning, as is not limited to the narrow "literal meanings" established by truth conditions.

5. It accounts for metaphorical as well as literal meanings.

6. It shows how the unconscious is involved in the creation of meaning.

7. It makes explicit a semiotic process of meaning-making, relates it to general synthetic cognitive functions, and thus should qualify as a model of a cognitive science.

---

[1] Of course, linguists of the formalist schools would merely dismiss such meanings as not part of semantics at all, but rather belonging to "import," "speaker's meaning," or "mere pragmatics." Obviously, it will have to be part of my argument that this kind of meaning is actually fundamental and primary, and that what they call semantic meaning is merely the unmarked, conventionalized shape of the sign--what's left after all the interaction with world, with intention, and with context is done with and it sits back on the shelf in the warehouse (lexicon), waiting for another chance to enter a molecule and participate in the dynamics of meaning.

[2] Actually, the case is probably more complex. More accurately, the name "Non-Aspirin" evokes a molecule {Aspirin / non-aspirin } which maps onto, by analogy, an already-familiar molecule of brand-name pain killers {Aspirin / Tylenol / Excedrine}. While this mapping might leave the identity of "Non-Aspirin" ambiguous (Tylenol or Excedrine?) a note on the box which reads, "Compare to extra strength Tylenol," disambiguates the analogy.

[3] One might compare this descriptive to that of the typical descriptive grammar, which would use the slot before the noun (or between the determiner and the noun) to define the positional category of Adjectival. Within this method, every term which could be given an adjectival form would be invited, at least theoretically, without regard to lexical content-- including a great many which would not be considered by any imaginable creator of text--"thirsty," "thirtieth," "boiled," "indexical," "foreign," and "illegal", to name a few. In a formal explanation of the sentence, such words would have to excluded by special lexical rules. If performance is given priority, on the other hand, only those semantic contents needed for the sentence would be entertained in the first place.