Title: COHESION AS LOGIC: THE POSSIBLE WORLDS OF MARVELL'S "TO HIS COY MISTRESS" ,  By: Karon, Jeffrey W., Style, 00394238, Spring93, Vol. 27, Issue 1
Database: MasterFILE Select

Cohesion analysis has found a comfortable home in literary stylistics. Though in recent years logical semantics has provided exciting new furniture for that home, logic and cohesion have not been productively united. Most accounts of cohesion mention so-called "logical relations," yet fail to explain the mechanism underlying many strong cohesive ties. I argue that by exploring the behavior of interclausal connectives, we can expand cohesion analysis in an useful direction; this expansion demonstrates how logic and language intertwine. I share with Thomas Pavel the conviction that "[t]he semantics of fiction has remained . . . at the periphery of critical attention. Yet a comprehensive theory of literature needs a viable account of literary content that would complement formal and rhetorical studies" (vii). Ultimately, logical semantics is crucial for literary criticism.

My project has four stages. First, drawing on M. A. K. Halliday and Ruqaiya Hasan's classic account of cohesion, I examine Andrew Marvell's "To His Coy Mistress" (see the appendix!. Second, I provide an argument that has been conspicuously absent from the literature on cohesion: not only does interclausal cohesion exist, but it is far from rare. For this argument, I focus on words many authors call "connectives." Third, the cohesive force of many interclausal connectives can be traced to their underlying logical structure. Fourth, though the data for my analysis is generated by our familiarity with Marvell's poem, the supplemented theory can illuminate a work whose critical ground is over trod.


Many stylisticians make use of cohesion analysis. They do not, however, agree on what constitutes cohesion. Simplifying schemes abound, sometimes purposely conflating the notions of cohesion and coherence. In a recent book, Sally Stoddard confines her attention to referential ties such as pronouns and articles after pointing out the vagueness in Halliday and Hasan's original definition (15). Regina Blass comments that "the concepts of cohesion and coherence sometimes seem to be almost as vague as the notions of text . . . and unity of meaning themselves" (15). When Gillian Brown and George Yule claim that "[lt]he cohesive relationship which particularly interests them [Halliday and Hasan] is that which they discuss under the headings reference, substitution, ellipsis and lexical relationships," they remain true to Halliday and Hasan's original scheme except for the odd omission of the important conjunction category (192). In a considerable simplification, Geoffrey N. Leech and Michael H. Short divide text cohesion into cross-reference (for example, the use of pronouns) and linkage. Apparently, linkage involves either "logical or other links between sentences" or "implicit connections of meaning" (79). In the category of linkage we would find examples of conjunction (which for Halliday and Hasan includes words such as yet, though, and therefore), but Leech and Short are not at a]l clear about the contrast between logical and "other links," let alone "implicit connections of meaning."

Different typologies of cohesive relations can be traced to the different purposes authors have when analyzing a text. Halliday and Hasan provide a wealth of tools for examining texts, but different authors select only what they view as important cases of cohesion for their analyses. In this I shall be no different. In particular, my examination of Marvell's poem will concentrate on referential, lexical, and especially conjunctive cohesion (Halliday and Hasan 31-87, 226-92). But actual confusion in the definition of cohesion may be traced to a failure to draw "the distinction between the `meaning relations' which hold between items in a text and the explicit expression of those `meaning relations' within a text" (Brown and Yule 195). That is, if certain words mark cohesive ties and if cohesion is a semantic relation as Halliday and Hasan assure us that it is (4), then the relation between text constituents can exist whether or not there is an explicit realization of it in the form of specific markers (Brown and Yule 195). Blass, on the other hand, believes that "cohesion is merely a surface symptom of some deeper relation which can exist independently of it" (17); the deeper relation requires that we move beyond even considerations of text coherence (74).

For the purposes of this analysis, I will side with those who believe that certain words function as cohesive ties because they signal underlying semantic relations. But this is to say little until we can characterize these relations in more depth. In the meantime, I note that there is one common assumption among those who use cohesion analysis: it is acceptable to study bonds between sentence clauses. As Leech and Short comment, studying sentence relations alone "seems rather restrictive for purposes of literary analysis" (256). In Halliday and Hasan's original work, cross-clausal cohesion can occur, as it does when a pronoun refers back to a referent introduced in the previous clause (8). But when clauses are connected by conjunctions such as but or therefore, the ties are syntactic rather than semantic, and thus are not cases of cohesion (8-9)

I will show that Leech and Short are correct in examining clausal relations as cases of cohesion. To do so involves isolating the mechanism that underlies a surprising amount of text cohesion, specifically a logical relation that can be the strong semantic relation of conditional logic. Thus, we shall be led from a supplemented theory of cohesion to an examination of the relation between language and logic. Interclausal and intersentential ties often exhibit the same type of cohesion, as I suspect stylisticians unconsciously realize. The proof, though, requires an analysis of logical structure.


I begin by applying an unsupplemented version of Halliday and Hasan's techniques. Marvell's "To His Coy Mistress" begins: "Had we but world enough, and time, / This coyness, lady, were no crime." Halliday and Hasan use the term "exophoric" to characterize pieces of text that refer to some aspect of the surrounding speech situation but that are not themselves cohesive (18). Without a prior reference, the first pronoun--"we"--may be functioning in this manner. Yet if we consider the title (which contains "his" and "mistress"), then "we" becomes anaphoric and thus cohesive. Likewise, "lady" is lexically cohesive, referring to "we" as well as "mistress." "This coyness" is more complex. It is anaphoric if we are willing to grant that "coyness" was introduced in the title. I would like to argue for the possibility that the phrase also functions cataphorically, that it also points forward in the text. Halliday and Hasan claim that a word such as this may be either anaphoric or cataphoric, but they do not tell us whether it could function both ways at once (68). Yet one of the interesting differences between natural language, whether or not in a literary context, and formal logic is the ability of a word or phrase in a natural language to perform this sort of double duty. A logically regimented representation of natural language, however, will assign single functions to syntactic units: for example, that of quantifier or variable. Though I claim that logic and cohesion are inextricably linked, I am not employing the notion of logic as a fully articulated, abstract system, for this would be to advert to the type of structures common in formal linguistics. The development of cohesion theory was, in part, a reaction against the employment of generative linguistics patterned after formal logic. Style analyses employing generative grammars focused on sentence structures rather than on links between sentences. Halliday and Hasan's work shows us that any linguistic approach confined to the sentence level is inadequate for the understanding of text.

In order to pick up the possible cataphoric reference, we need to move forward through the poem to lines 27-28: "My echoing song; then worms shall try / That long preserved virginity;" (emphasis added). We often use this to signify nearness, whereas that signifies something not quite so near (Halliday and Hasan 57). The latter demonstrative indicates a spatial or temporal lapse between the introduction of the topic of the mistress's coyness and its reintroduction as a reference to her virginity. While "yonder all before us lie" (line 23) invokes spatial distance, "Deserts of vast eternity" (line 24), "echoing song," and "And your quaint honour turn to dust" (line 29) separates the cohesive references by the inexorable march of time. The phrase "your quaint honor" points back to the original "coyness," just as "that long preserved virginity" also points back after a long passage that sets out a possible future.

If we focus only on the first two lines of the poem, then we have exhausted the resources of Halliday and Hasan's account. As we read on we find that after its first appearance, "time" becomes highly cohesive. The word appears once in both the second and third stanzas. In the second, time is anthropomorphized ("Time's winged chariot"), whereas in the third it is personalized ("our time"). The possessive "our" in "our time" signals empowerment and accords with the narrator's wish to grasp the moment. (Thus, I think that we should place the poem back into the carpe diem tradition, rather than side with critics who believe that Marvell is in some way subverting the form. In a later section, I consider whether there is evidence for their claim.) On the way to developing this personally owned time, Marvell employs lexically cohesive variants such as "more slow" (referring to how fast the narrator's love should grow) and "vast eternity" (mentioned above). Most notable is the narrator's recitation of a mathematical series, assigning each number to the contemplation of a different feature of his love. He begins with "ten years before the Flood," moves on to "An hundred" (line 13), "two hundred" (line 15), "thirty thousand" (line 16), and finally to "an age" (line 17) and "the last age" (line 18). Thus, the whole series refers to time, with the definite article "the" used as a way to bring the series to a close or limit. According to Halliday and Hasan, the usually signals anaphoric reference, pointing backward (72-73). The definite article also indicates specificity (70-71), which here fits with Marvell's need to indicate the end of an infinite sequence. Line 19--"For, lady, you deserve this state"--would make sense only if there were a final state or limit to the sequence. As with "this coyness," the use of "this" in "this state" can indicate a present fact. But unlike the fact of his lady's coyness, this "fact" is introduced within the scope of a hypothetical conditional.

If we were to follow Halliday and Hasan, we would find no further instances of cohesion between the poem's first two lines. Instead, they would argue that the lines are linked by structural relations, in particular, some form of coordination (this category includes what others call "subordination" and "correlation"). But these structural relations are not cohesive.

For the moment, let us follow Leech and Short and consider the possibility of some other cohesive tie between the first and second lines. What are the semantic relations that link one with another? As a first pass, we can see that Marvell presents us with a conditional, often expressed in the form "If A, then B." Indeed, many traditional analyses note this fact, often referring to the poem's form as a "hypothetical syllogism." Such a conditional statement links one state of affairs with another, though the strength of this link may vary considerably from mere conjunction to full-blown causation. Few commentators have noted, however, that Marvell's conditional is special: it has acounterfactual antecedent, signaled by the verb "had." The resulting subjunctive conditional is not quite captured by calling it a hypothetical syllogism. The use of the verb "had" implies that the lovers do not have "world enough, and time"; but if they did, then there would be no problem with the lady's coyness. The consequent, by asserting that in the context of the antecedent condition the lady's coyness would not be a crime, implies that it is a crime. The rest of the stanza (lines 3-20) works out the contours of the world created by the antecedent, a world where the lovers have unlimited time.


I shall show that we should analyze interclausal connectives such as but, yet, and therefore in terms of logic, and logic in terms of conditional connections. These connections can be understood as representing ties between possible worlds. In Halliday and Hasan's view, the connectives mentioned are examples of conjunctive cohesion only when they connect sentences (237-38,257). Yet whatever the underlying semantics are for these words when they reach across sentences, it is plausibly the same when they connect clauses. Thus we are correct to search for interclausal cohesion; our search, though, leads us to characterize the underlying semantic structure of a whole class of cohesive relations: that is, cohesion reduces to logic in these cases.

At the end of the last section, I used the word world loosely but appropriately. In recent years, Pavel, Marie-Laure Ryan, and many others have tried to apply the notion of a "possible world" to the analysis of literary texts, a notion taken over from the seminal works of philosopher-logicians such as David Lewis and Saul Kripke. In the context of the present paper, a possible-worlds approach offers an alternative to the unsatisfactory truth-functional analysis of "If . . . then" statements. According to a truth-functional analysis, if the antecedent of a conditional is false, then the whole conditional statement must be true, no matter what the consequent may be. Since subjunctive conditionals often have contrary-to-fact antecedents (as with Marvell's first line), all of them would be true according to a truth-functional analysis. There would be no way to distinguish true from false ones, or likely from unlikely ones. Thus some logicians invoke the notion of possible worlds as a way to explain the semantics of conditionals and subjunctive conditionals in particular. Where they have differed considerably is in how exactly to construe possible worlds themselves. According to David Lewis--in his often cited Counterfactuals as well as the more recent On the Plurality of Worlds--we should view possible worlds as real, distinguishing them from the actual one in which we three-dimensional creatures breathe. For Saul Kripke, on the other hand, possible worlds are constructs of some sort from the actual world: we stipulate their existence without thereby being committed to some bizarre ontology of strange entities. Possible worlds are not the sorts of things that can be viewed through special telescopes (44): that is, possible worlds are created, not discovered. Between these extremes are many variants, but with Simon Beck we could also identify a third view, that of "antirealism" in which possible worlds do not exist: they provide only "a convenient manner of speaking" (121).

Recent attempts to employ possible worlds in literary analysis settle about the first or second view. In a lively new book, Marie-Laure Ryan decides that Lewis's theory "offers a much more accurate explanation of the way we relate to these worlds" (21). Thus she is willing to accept the full ontology of these new entities. Pavel, on the other hand, objects that Lewis "has defended the view that all possible worlds, together with all the objects that populate them. are as real as our own world. But this form of possibilism is an extreme position, which offends our most common intuitions" (49). Relying on Kripke's definition of a model structure (44), Pavel implies that we should consider possible worlds as "actual abstract entities or as conceptual constructions" (49). Umberto Eco, though rejecting the "realistic approach" of Lewis, views possible worlds as "cultural constructs, matter of stipulation or semiotic products" (343).

But as Wolfgang Heydrich points out, those who have taken up possible worlds into literary analysis may not have recognized "that wherever the very nature of possible worlds was under discussion there was (and is) more quarrel and controversy among philosophers than agreement and consensus" (190). One problem with Lewis's theory was alluded to by Pavel: we may be uncomfortable with an explosion of new, mysterious entities into our ontology. And in Pavel's view this leads to the absurd position that fictional worlds exist independently of the authors who write about them (49). For Ryan, though, the proposed "realness" of these worlds is what recommends the theory, since readers relate to fictional worlds as though they are real (21), a sentiment echoed in Lubomir Dolezel's suggestion that "possible worlds acquire fictional existence by being discovered' (235). Writers such as Pavel and Ryan do not deal with other equally serious problems raised by philosophers. As W. V. Quine puts it,

[w]hen modal logic has been paraphrased in terms of such notions as possible world or rigid designator, where the displaced fog settles is on the question when to identify objects between worlds, or when to treat a designator as rigid.... (Theories 174)

That is, in Quine's view Lewis never provides a problem-free account of crossworld identity of individuals, an account crucial for his theory. Kripke's theory, on the other hand, avoids ontological multiplication by offering a controversial theory of reference that depends on the notion of a "rigid designator." Yet as Alan Sidelle points out, this concept has its own metaphysical baggage (410-11). Pavel, recognizing that there are problems involved in adverting to possible worlds, advises us that "the notion of world as an ontological metaphor for fiction remains too appealing to be dismissed.... An attempt should be made at relaxing and qualifying this crucial notion" (50). We should understand possible worlds "as abstract collections of states of affairs" (50).

Perhaps John Pollock is too hopeful in believing "that ontological questions about possible worlds can be safely separated from the question how to analyze subjunctive conditionals" (15). Rather than settle the question of ontological status here, I prefer to opt for a more austere theory of the sort constantly championed by Quine: multiply your entities as slowly as possible ( Ways of Paradox 264). For example, I see no explanatory value in analyzing possible worlds into states of affairs, as many do, since these seem to me to be alternative ways of saying the same thing. Thus in order to analyze Marvell's poem, we need to think of a conditional as a link between possible states of affairs (worlds), some of which actually obtain. Whatever else, different possible-world analyses share the assumption that thinking in terms of possible states of affairs is natural. Recent evidence reported by James P. Brynes on children's mastery of hypothetical constructions tends to bear this out.

Where has this detour into logical semantics gotten us? For one thing, we can see that coordinating and subordinating conjunctions (as Halliday and Hasan use the term) may signal the strength of logical connections. To use one of Halliday and Hasan's examples (252), surely in "The total came out all wrong, although all the figures were correct," there is a connection between two facts or states of affairs, one they call "contrastive." Following Leech and Short it is difficult to see why it would not be an example of cohesion. In "The total will come out all right, unless the figures are incorrect," "unless" appears to signal full-strength conditionality. I suspect that all the contrastives, such as but and yet, display subtle shadings in the strength of connections they signify. Although authors such as Diane Blakemore rightly claim that the semantics for but cannot be captured in a truth-functional analysis (125-41), they have trouble providing a clear mechanism that underlies its cohesive force. At this point, we can see what motivates the selection of certain interclausal conjunctions over others: the strength of the connection between states of affairs picked out by different clauses can vary. According to this view, there is no absolute distinction between "loose" conjunctive links and "tight" causal conditional links: the distinction is a matter of degree.

My suggestion here is somewhat prefigured by Teun A. van Dijk's analysis of connectives. As he notes,

The typical task of connectives is to express relations between facts. These relations may be very loose, as in conjunction and disjunction, or they may have a stronger character, in the sense that facts may somehow DETERMINE or CONDITION each other. The large class of different types of connectives expressing these DEPENDENCY relations between propositions or facts, will be called CONDITIONALS. (67)

Van Dijk casts his formulations in terms of possible worlds:

With the exception of enumerative conjunction [including what he calls the "neutral" and] and disjunction, natural connectives are of the CONDITIONAL types in the sense that the consequent is to be interpreted in worlds determined by the antecedent. . . (89)

Although pursuit of a possible-worlds analysis lays him open to the same attack leveled at others who have not accounted for the ontological status of these entities, I think that he is correct in presenting natural language connectives as ranging over a "scale," leading up to some strict form of implication between clauses (67). Van Dijk also notes that--with the exception of the neutral and--all the "conjunctive" connectives "condition" one clause or the other (89). No textbook on logic seriously questions the central role that the conditional--however interpreted--plays in our reasoning. At least, then, for the large class of cohesive elements that Halliday and Hasan lump under "conjunction," cohesion consists of logic. That is, interclausal connectives can perform the same duty as intersentential connectives. Their semantic function is to relate states of affairs, otherwise known as possible worlds. Although Leech and Short believe that only a connective such as therefore signals the strongly cohesive relation of logical reason (250-51), we need to realize that conditionality itself comes in degrees. This is why van Dijk rightly refuses to separate rigidly the "classes of natural connectives" (67).


As I said above, Halliday and Hasan would not recognize any cohesion between Marvell's first two lines. They would recognize a strongly cohesive element at the beginning of the second stanza, the contrastive "But": "But at my back I always hear / Time's winged chariot hurrying near." What is the contrast with, though? It must be with the conditional world set up in the poem's first line and worked out in the rest of the first stanza: the world of unlimited time. If the poem began with an indicative conditional (in which the antecedent is not contrary-to-fact), then Marvell would be guilty of an elementary logical mistake: namely, denying the antecedent. Indeed, most commentators assume that there is an obvious fallacy. They differ over what follows from this supposed fact. Anne E. Berthoff, though noting the poem's "elegant logic" and "the logical pattern of the argument," nonetheless places it in the carpe diem tradition (111-13). According to B. J. Sokol, appreciating the poem's tight logical structure (including its subjunctive conditional) leads us to the conclusion "that Marvell deliberately misuses logic, and so uses illogic to a purpose" (247). The fallacy forces the reader to view the poem as an attack on the whole carpe diem tradition. Margarita Stocker claims that Marvell "is unconcerned to maintain more than the generic convention of [the] syllogism" in order to "subvert" the logical apparatus in some less obvious manner (343n3). And according to Bruce King, though the poem has a "highly articulated logic," it is the allegorical dimensions that undermine the carpe diem tradition (66,74-76). Contrary to those who believe that the logical structure is either unimportant or there to be subverted, I argue that the poem's rhetorical force is intimately connected with its subjunctive conditional structure, which at this point we can recognize as bound up with the poem's tight cohesion. I agree with Berthoff that the poem has an "elegant logic"; in fact, we should question whether the poem's reasoning really is flawed.

This is not an easy question to answer, precisely because we are dealing with a subjunctive conditional that connects possible states of affairs. The stronger the connection between the antecedent and consequent, the more difficulty a reader will have in breaking the link when the poem's speaker "denies the antecedent." The consequent also will fail. Of course, to say "a reader will have a difficult time" is a statement of the poem's rhetorical force and at first blush may seem not to bear on its logic. But consider someone saying, "If you push that door, it will open; but if you don't push it, then it won't open." Is there an "obvious" fallacy here, even though the speaker denies the antecedent? If the reasoning appears unobjectionable, it is because we can supply a context in which we accept a number of implicit additional conditions: for example, that no one else is around to open the door, that it is latched securely, and so on. The subjunctive conditional functions smoothly because we easily can recognize the contours of the appropriate state of affairs or possible world. Below I argue that Marvell's use of the subjunctive conditional keys us to appreciate the contours of another possible world, a world possessing a particular character that supplies some of those additional implicit conditions accounting for the poem's logical and rhetorical force. The poem's logic is not so obviously "flawed."

David H. Sanford recounts evidence that people tend to reason from conditionals to the associated biconditional (237). A biconditional (often represented by logicians as "A if and only if B") can be analyzed as the conjunction of two conditionals: "If A, then B" and "If B, then A." Marvell links the states of affairs so strongly that we may well be responding to the subjunctive conditional as though it were a biconditional; if the poem really does present a biconditional, then the fallacy disappears completely even in standard truth-functional logic. This would not mean that we or his mistress must accept the poem's conclusion. After all, we still can dispute the appropriateness of conjoining certain possible worlds with others; in other words, we can deny the truth of the whole conditional statement. The speaker's mistress, in particular, might object that the state of her coyness is not relevant to--which is to say that it should not be connected to--the fleeting nature of time. I am not sure how we could show that Marvell actually meant us to think of the possible worlds as being connected biconditionally. On the other hand, I cannot see how to show that he did not. My point is that just two of the possible approaches to the logic of conditionals provides us with reason to doubt that the poem's "logic" and "rhetorical force" should ever have been split from each other. When Stocker claims that "Marvel! is unconcerned to maintain more than the generic convention of [the] syllogism" (343n3), she unfairly relegates logic to a far corner of the literary landscape. I want to revive the belief that the poem derives a good part of its rhetorical force from logic yet deny that there must be a horrendous fallacy at its heart.

The common intuition that there are explicit logical links in the poem is heightened by Marvell's beginning the third stanza with "Now therefore"; in fact, the last sentence of the stanza (and poem) begins "Thus." No doubt Leech and Short are correct when they remark that therefore signals a relation of reason (250), but what is the function of "now"? In this case, the word is not functioning as a demonstrative, but rather as what Halliday and Hasan call a "continuative." Such a use signals "a new incident in the story, a new point in the argument, a new role or attitude being taken by the speaker, and so on" (286). Indeed, the rest of the stanza reads not as a cold conclusion to an argument, but a call to vividly amorous activity.

The last two lines begin with a strongly cohesive element: "Thus, though we cannot make our sun / Stand still, yet we will make him run." "Thus" connects us not to the immediately preceding sentence (or in fact to the rest of the stanza), but rather to the actual world set out in the second stanza: that is, the world of fleeting time and lost opportunities. Is the "though" performing a contrastive role as "although" did in Halliday and Hasan's example, "The total came out all wrong, although all the figures were correct"? No doubt it is, in addition to a stronger logical one; this role can be indicated by rewriting the line as "Even if we cannot make our sun stand still, we can...." The "even" seems to modify the conditional "if . . . then" in a complex manner, implying at least the truth of the antecedent (this possibility hardly exhausts all the semantic overtones). "Though" behaves in very much the same manner and is another example of a cohesive tie that works on the basis of conditional connections: semantic relations that are the basis for what we loosely refer to as "logical relations."


Sticking closely to Halliday and Hasan's view of cohesive linkage impoverishes our ability to characterize the semantic texture of the poem quite aside from any use of vivid or stirring language. Availing ourselves of a richer concept of cohesion, though, we can describe Marvell as setting up a powerful counterfactual world, the contours of which are described in the first stanza.

He then destroys this world in the second stanza, invoking images of destruction and decay, picturing what Cleanth Brooks calls a "bleak reality" (162). We are left with the denial of the original consequent (line 2): the third stanza presents a possible world of cavorting, lustful life. If we do not recognize the logical connections inside sentences as being crucial for cohesion, we cannot make sense of the compelling nature of the poem. Unsupplemented, Halliday and Hasan's analysis does help us understand features of the poem's individual worlds For example, the use of "now" to introduce the third stanza tells us that though so far only possible, its world can be actualized immediately. A supplemented account of cohesion, however, links these separate worlds together.

Though contemporary critics have downplayed the role of the "syllogism" in favor of investigating Biblical or philosophical references, we find that they have an implicit awareness of the logical structure, as when Stocker comments,

This vision of lengthy courtship is the unavailable, and undesired, model of their future (or "prophecy"). . The first act of this futurist projection in the poem is the first paragraph's mocking vision, here, of a hypothetical courtship . . . . (207-08)

I wholly approve of her use of words such as "model" and "hypothetical"; Stocker just fails to see that she is apprehending the poem's logical structure. But this structure is not a bare skeleton from which ornamentation hangs; rather, it is formed by complex, cohesive semantic ties.

Failure to appreciate logical structure can lead critics far astray. Balachandra Rajan claims,

"To His Coy Mistress" is probably Marvell's most destructive poem. Its strength is that having turned against itself in the expected manner of ironic poems, it then turns against its own internal objections, leaving us with the desert that is the poem's centre. It is inconclusive because of its consistent, subversive energy Marvell's other poems do not demolish themselves as thoroughly . . . . (163)

Rajan focuses on Marvell's negative imagery without placing those images in their proper sequence. As I indicated above, Marvell directs "subversive energy" against one possible world in order to destroy it: that is, to deny its actuality. The image of the desert may lie at the poem's literal (graphological) center, but hardly at a thematic one. To cast the entire poem as destructive is as much a failure of logical analysis as of literary criticism.


Leech and Short suggest a "scale of cohesiveness" on which therefore would be the strongest link, and and the weakest (250-51). They are on the right track, but the real dirty work is figuring out what comes in between. I have tried to show that connectives often signal weak or strong conditional connections, and these in turn underlie logical relations between text constituents. While investigating these connectives, we find no reason to claim that cohesive linkage between clauses is different in kind from the linkage between sentences. In the case of "To His Coy Mistress," failing to examine the logical links between clauses would prevent our seeing how the poem unfolds within the confines set up in the first two lines.

Although generally enthusiastic in his review of Halliday and Hasan's Cohesion in English, Stephen A. Bernhardt saw their book as a first step. Future work would

demand the development of more delicate models of analysis which account for the relative force of various ties and the effects of deliberate versus incidental cohesion. (50)

For at least two reasons, the literature on cohesion contributes little toward characterizing the "relative force" of ties. First, too many authors lump conjuncts such as but, yet, or and together, then contrast them with the "strongest" link such as therefore. This in turn can lead to a dichotomy like Winifred Crombie's in which semantic relations are either "associative" or "logico-deductive" ( I 13). But as I have argued, these cohesive words lie on a semantic continuum. Second, no one has tried to isolate the mechanism at work in the cohesion signaled by connectives; authors assume that cohesion itself is the most basic mechanism. Instead, I have analyzed a large class of cohesive devices in terms of conditional connections and hence in terms of logic. The conditional connections, in turn, are to be understood by reference to possible worlds, or states of affairs. Although some readers may find this scarcely less puzzling, I think that the arrangement has obvious intuitive advantages. As Robert C. Stalnaker replies to those who find references to possible worlds completely mystifying,

[a]n analysis makes a claim about a relation among concepts which, if accepted, can be informative in either direction, or in both directions. It may be as helpful in explaining an obscure concept to reduce other things to it as to reduce it to other things. (54)

In addition, we can see what so many connectives have in common, and we can begin to provide a "scale of cohesiveness." As far as determining when "deliberate" cohesion occurs, a supplemented account assumes that speakers or writers choose particular connectives precisely because they signal differences in the semantic ties between states of affairs.

Perhaps no one will account satisfactorily for possible worlds. Certainly I have not argued for any particular interpretation; instead, I concentrate on the primacy of the conditional, rather than the nature of possible worlds. I am sure that even if possible worlds turn out to be nothing more than a "convenient manner of speaking," conditionality will continue to figure prominently in logical semantics. Besides, when we worry that possible worlds may be nothing more than metaphors, we forget that the language of logical analysis--filled with statements about premises supporting conclusions and conclusions following premises--is filled with metaphor. But this hardly requires that such language forfeit its place in discussions of reasoning.

Our ardor for linguistic theory has cooled. No longer do we believe that a straightforward application of some new formalism will spin out literary insights. Yet perhaps Halliday and Hasan are too modest when they claim that "[t]he analysis of cohesion .. . will not in general add anything new to the interpretation of a text" (328). The study of cohesion firmly embeds logic in literary criticism. My program for revising cohesion theory, beginning as it does with our intuitions of the "logic" in Marvell's poem, ultimately is tested by turning it back on the poem. By doing so, we find that the old divisions between logic and language and between logic and rhetoric should never have occurred.

To His Coy Mistress
Had we but world enough. and time,
This coyness, Lady, were no crime.
We would sit down, and think which way
To walk, and pass our long love's day.
Though by the Indian Ganges' side
Should'st rubies find: I by the tide
Of Humber would complain. I would
Love you ten years before the flood:
And you should, if you please, refuse
Till the conversion of the Jews.
My vegetable love should grow
Vaster than empires, and more slow.
An hundred years would go to praise
Thine eyes, and on thy forehead gaze.
Two hundred to adore each breast:
But thirty thousand to the rest.
An age at least to every part,
And the last age should show your heart.
For, Lady, you deserve this state;
Nor would I love at lower rate.

    But at my back I always hear
Time's winged chariot hurrying near:
And yonder all before us lie
Deserts of vast eternity.
Thy beauty shall no more be found;

Nor, in thy marble vault, shall sound
My echoing song: then worms shall try
That long preserved virginity:
And your quaint honour turn to dust;
And into ashes all my lust.
The grave's a fine and private place,
But none, I think, do there embrace.

     Now therefore, while the youthful glew
Sits on thy skin like morning dew,
And while thy willing soul transpires
At every pore with instant fires,
Now let us sport us while we may;
And now, like am'rous birds of prey,
Rather at once our time devour,
Than languish in his slow-chapped power.
Let us roll all our strength, and all
Our sweetness, up into one ball:
And tear our pleasures with rough strife,
Through the iron gates of life.
Thus, though we cannot make our sun
Stand still, yet we will make him run.

Works Cited

Allen, Sture. Possible Worlds in Humanities, Arts and Sciences: Proceedings of Nobel Symposium! 65. New York: Gruyter, 1989.

Beck, Simon. "The Method of Possible Worlds." Metaphilosophy 23.1-2 (1992): 11931.

Bernhardt, Stephen A. Rev. of Cohesion in English, by M. A. K. Halliday and Ruqaiya Hasan. Style 14 (1980): 47-50.

Berthoff, Ann E. The Resoled Soul A Study of Marvell's Major Poems. Princeton: Princeton UP, 1970.

Blakemore, Diane. Semantic Constraints on Relevance. Oxford: Blackwell, 1987.

Blass, Regina. Relevance Relations in Discourse. Cambridge: Cambridge UP, 1990.

Brooks, Cleanth. "Andrew Marvell: Puritan Austerity with Classical Grace." Andrew Marvell. Ed. Harold Bloom. New York: Chelsea, 1989. 161-68.

Brown, Gillian and George Yule. Discourse Analysis. Cambridge: Cambridge UP, 1983.

Brynes, James P. "Acquisition and Development of If and Because Conceptual and Linguistic Aspects." Perspectives on Language and Thought: Interrelations in Development. Ed. James P. Byrnes and Susan A. Gelman. Cambridge: Cambridge UP, 1991. 354-93.

Crombie, Winifred. "Semantic Relational Structuring in Milton's Areopagitica." Language, Discourse and Literature: An Introductory Reader in Discourse Stylistics. Ed. Ronald Carter and Paul Simpson. London: Unwin, 1989. 113-20.

Dijk, Teun A. van. Text and Context: Explorations in the Semantics and Pragmatics of Discourse. New York: Longman, 1977.

Dolezel, Lubomir. "Possible Worlds and Literary Fictions." Allen 221-42.

Eco, Umberto. "Report on Session 3: Literature and Arts." Allen 343-55.

Halliday, M. A. K., and Ruqaiya Hasan. Cohesion in English. London: Longman. 1976.

Heydrich, Wolfgang. "Possible Worlds and Enkvist's Worlds." Allen 187-98.

King, Bruce. Marvell's Allegorical Poetry. New York: Oleander, 1977.

Kripke, Saul A. Naming and Necessity. Cambridge: Harvard UP, 1980.

Leech, Geoffrey N., and Michael H. Short. Style in Fiction: A Linguistic Introduction to English Fictional Prose. New York: Longman, 1981.

Lewis, David. Counterfactuals. Cambridge: Harvard UP, 1975.

--------. On the Plurality of Worlds. New York: Blackwell, 1986.

Marvell, Andrew. "To His Coy Mistress." Selected Poetry and Prose. Ed. Robert Wilcher. New York: Methuen, 1986. 40-42.

Pavel, Thomas G. Fictional Worlds. Cambridge: Harvard UP, 1986.

Pollock, John. Subjunctive Reasoning. Dordreicht: Reidel, 1976.

Quine, W. V. Theories and Things. Cambridge: Harvard UP, 1981.

--------. The Ways of Paradox and Other Essays. Cambridge: Harvard UP, 1976.

Rajan, Balachandra. "Andrew Marvell: The Aesthetics of Inconclusiveness." Approaches to Marvell: The York Tercentenary Lectures. Ed. C. A. Patrides. London: Routledge, 1978. 155-73.

Ryan, Marie-Laurel Possible Worlds, Artificial Intelligence, and Narrative Theory. Bloomington: Indiana UP, 1991.

Sanford, David H. If P, then Q: Conditionals and the Foundations of Reasoning. London: Routledge, 1989.

Sidelle, Alan. "Rigidity, Ontology, and Semantic Structure." The Journal of Philosophy 89 (1992): 410-30.

Sokol, B. J. "Logic and Illogic in Marvell's 'To His Coy Mistress.'" English Studies 71 (1990): 244-52

Stalnaker, Robert C. Inquiry. London: MIT P, 1989.

Stocker, Margarita. Apocalyptic Marvell. Athens: Ohio UP, 1986.

Stoddard, Sally. Text and Texture. Patterns of Cohesion. Norwood: Ablex, 1991.


By Jeffrey W. Karon, University of South Florida