|AppliedPhilosophy| |Courses| |Bibliography| |Biography| |Papers|









Published in:





Rastko Močnik & Aldo Milohnić (eds.),

Along the Margins of Humanities. Seminar in Epistmology of the Humanities, Ljubljana: Institutum Studiorum Humanitatis, 1996, pp. 103-136



Family Disturbances

Metaphors, Similes and the Role of 'Like'


A study on Continuity and Inreducibility


Comparisons, Similes, and Metaphors


"Virtual reality is wholy, totally real, but not!"

"Short cuts", R. Altman


The relationship between metaphors and similes has been the main topic of theories of metaphor in philosophy of language, linguistics and theory of science for the three decades. The purpose of the present paper is to give reasons not included in the non-comparativist metaphor theories for the assumption that metaphors, though apparently closely related to simple predicative similes of the form 'X is like Y', are neither reducible to similes nor semantically explicable by them because these two tropes, while expressing different properties of the objects related, i. e., conveying different semantic contents ('being something' and 'being like something'), serve different linguistic functions. The special concern of the paper is to uncover the ambiguous effects of the semantic marker 'like' in both literal and figurative environments. I will show that irreducible ontological presuppositions, entailed by the semantic marker of similarity, preclude a continuous transition from similes to metaphors and require that the relationship between the two tropes be subverted such that metaphors are to be viewed as logically antecendent to similes. As a consequence, the similarity or likeness implication between things, though not irrelevant to the general understanding of metaphors, is not constitutive of their linguistic function and their meaning. Instead, metaphors appear to be more closely related to the structure of literal assertions than to that of figurative comparisons.


1. Confusions About Tropes and the Similarity Implication

One of the most discussed traditional assumption about metaphors is that they are elliptical similes because they, like similes, convey a figuratively expressed similarity between unrelated objects without, however, using any explicit semantic marker of similarity. Accordingly, for some authors similes are figurative comparisons with an explicit semantic marker of likeness (such as "is like", "similar to", "as", "as well/many/much/ as" etc.) while metaphors are similes with an implicit semantic marker. This entails two important points concerning metaphors and similes: first, that they have the same semantic content despite the use of different linguistic means, and second, that they belong together due to their family resemblance which consists in sharing a figurative comparison while other related tropes, such as comparisons, analogies and models, convey a literally intended similarity.

Although many theorists of metaphor since the 1954 appearance of Max Black's revolutionary essay "Metaphor" (Black 1962) have been sceptical of these assumptions, especially of the latter one concerning similarity as the common linguistic function of both similes and metaphors, the reasons for this scepticism are not strong enough to preclude the possibility that similarity or likeness is in some way involved as a common implication of both tropes. For to imply likeness between things and, on this basis, a reducibility of similes to metaphors, and vice versa, is not to assume equality or identity of the things related, as has been repeatedly asserted.[1] Rather the scepticism about mutual reducibility between metaphor and simile should be founded upon the fact that metaphors and similes have a different linguistic form and that, by this very fact, they indicate different linguistic functions. Adding a semantic marker is not trivial, as has been claimed against the reductivist theories of metaphor.[2] For, as we intuitively grasp, it is not the same thing to say that something is something and that something is like something. Moreover, we presume that it is not only the difference in the grammatical form between those sentences built upon predications and those containing resemblance indicators which calls attention to the difference between similes and metaphors, but also that there must be a different linguistic function and a different logical basis underlying the predication and comparison markers respectively.

Thus confusions about tropes result as a consequence of non-obvious assumptions about what exactly is implied in the thesis of mutual reducibility between similes and metaphors, and the front lines between different authors are anything but clear. As D. Davidson rightly suggests[3], the conflation of similes and metaphors and the confusion about the role of similarity in both tropes derive from the wrong assumption that the "abbreviation" of a simile by a metaphor implies its reducibility to metaphor. If this were the case, he says, we could not account for the linguistic difference between the tropes at issue, but he does not explore or elaborate upon the differences in the linguistic form which seem indispensable for a just account of metaphor. Nevertheless, one can add to Davidson's observation that in speaking of mutual reducibility between metaphors and similes we, as a matter of fact, speak of two opposite transformation procedures within grammar, i. e., of abbreviation (of similes) and of extendability (of metaphors). For by adding to a metaphor of the form "X is (a) Y" the semantic marker of similarity, we extend the expression grammatically; conversely, by omitting the semantic marker 'like' in a simile of the form "X is like (a) Y", we simply reduce the simile grammatically. Traditionally, however, when speaking of reduction, a semantic reducibility between the tropes is usually intended and not just a grammatical one.

But the very fact that we are primarily dealing with grammatical ("merely linguistic") transformations indicates that a further examination of the issue is needed. For if we say that metaphors are grammatically abbreviated similes and that similes are grammatically extended metaphors—and we still intend by this a semantic reducibility—we strongly suggest that (a) the difference between the two tropes is merely a grammatical one, while (b) their semantic identity remains untouched by grammatical changes, and (c) the grammatical interchangeability is only justifiable by virtue of the semantic identity, and d) the relationship between metaphors and similes is grammatically and semantically one of symmetry—meaning that every simile is reducible to the corresponding metaphor and every metaphor is extendable to its related simile.

Obviously, the latter implication involves the former three and it is exactly this implication which contradicts our intuition about metaphors and similes. For the intuition is incompatible with the idea that every metaphor can be extended and replaced with a (grammatically) corresponding simile without any significant change. Hence, instead of speaking of reduction, abbreviation, and extension it seems better to speak, though somewhat vaguely, of transformation—making room for the possibility that both the grammatical form and the meaning of the expressions undergo some kind of complex change. By focusing on the possiblity of transformation, we do not merely restrict ourselves to a blind intuition about metaphors, but, instead, we allow for a comprehensive analysis. This means that we (a) acknowledge the factual, material or grammatical difference between similes and metaphors, (b) assume that the superficial or material difference in grammatical form is due to an already existing difference in meaning, i. e., (c) allow that metaphors and similes are not interchangeable but rather at odds with one another, and (d) expect that the change in grammar entails a difference in meaning.

In stating this it is, however, not also asserted that metaphors have no relation to similarity, upon which all comparison expressions with explicit semantic markers rely. Rather there is the implication that similes are not immediately reducible to metaphors and that metaphors are not immediately extendable to similes unless a semantic shift or change of meaning within an expression is involved.[4]

A further source of confusion about metaphors is the assumption, which, though already discarded, has still not been sufficiently clarified, namely that all metaphors are not only grammatically and semantically reducible to similes, but are also resolvable into literal comparison statements, thereby providing a literal interpretation of the corresponding simile. This traditional account of metaphor has been labelled comparativist metaphor theory and rejected by Max Black (1962). But, as a matter of fact, there is a more moderate comparativist position, called nonreductive comparativism *(profoundly criticized by L. Tirrell, cf. Tirrell 1991), which assumes that there is a continual transition from similes to metaphors, and viceversa, but which does not maintain that there is an immediate passage from literal comparison statements to figurative expressions or tropes (Fogelin 1994). The latter assumption—called "Black's error" (cf. Fogelin 1994)— about the relationship between figurative tropes such as similes and metaphors and literal comparison statements cannot, as a matter of fact, be ascribed even to such traditionalist accounts of metaphor as Aristotle's. For, as Fogelin correctly assumes, although Aristotle asserts that metaphors and similes call attention to astonishing likenesses between unrelated things, and even though he asserts that metaphors belong to similes, he does not assert that the similarities suggested by similes and metaphors are as such reducible to comparisons—understood as "already existing similarities between things" and expressed by literal comparison statements. Quite on the contrary, Aristotle speaks of "congenial" and "astonishing" similarities just as modern writers on metaphors claim for. (*Thus Aristotle, while classifying metaphors as "a kind of comparison", is, if not creativistic in interpreting metaphors, then at least vague about the status of "novel" and "insight providing" metaphors. For as a matter of fact, what he does is explaining—but not reducing or replacing—metaphors through similes within the particular, namely epistemological discourse.* 12.1.1997)

However, a more important fact about the history of metaphor theories than Black's error itself seems to be that no "enemy of metaphor" (Black) has been able to explain away the notorious rhetorical surplus of this trope as opposed to its commonly acknowledged cognitive value[5], and no "friend of metaphor" has been able to eliminate the comparison-based implication of likeness in metaphors from their linguistic form and, thereby, to radically distinguish metaphors from similes. This seems to be the reason why different theorists of metaphor mutually reproach each other with bad comparativism.(*Indurkhya on Black; Tirrell*)

Given this historical background it is certainly nothing revolutionary to say that the translatability of metaphors into literal expressions should not be considered a necessary consequence of the apparently very close grammatical and semantic relationship between metaphors and similes ("figurative comparisons"). Nor should it be startling to maintain that the similarity implication we feel to be essential to metaphors does not necessarily entail the semantic reducibility of metaphoric expressions to statements of similarity. But it is certainly a small revolution within theory to say, as Davidson does, that metaphors begin with and live from nothing but the literal meaning of words. There seems to be reason enough to take this idea for a crucial point about metaphors without being obligated to accept several other assumptions of Davidson's which are central to his position.

Indeed, Davidson's thesis is unique in the sense that it, contrary to excessive metaphor theories, gives metaphors as small a semantic credit as possible while leaving room for as great a meaning-effect as possible. If we, along with many authors, attempt to explain novel metaphors as conveyers of new similarity aspects (Black), cognitive contents (Indurkhya) and also of new linguistic meaning (Kittay), then we must, as I believe, begin with Davidson's general thesis that there is no metaphoric meaning of words apart of their literal meaning, *and this means more precisely: antecendent to their usage in a metaphoric framework*. For what is authentically novel in a language is itself already post-metaphorical; *the label "metaphoric" in dictionary entries indicates what has been called "dead" or unfresh metaphors*. So language can only be creative of meanings and cognitive contents on the basis of metaphoric procedures *regardless of the ultimate status of the literal language itself, i. e. of how it "fits" with reality*. Metaphors occur only within, and by means of, the literal language. In other words, metaphors seem to be possible in a language only if "everything" in language is not already a metaphor, *though the language itself might rely upon a "metaphoric" conceptual apparatus (Lakoff/Johnson 1981)*.

If this is true of language and metaphors, then any account of how these tropes "work" (Black), how they arise in language, must entail an account of literal linguistic procedures or, at least, be related to such an account (Kittay 1987; also Haack 1994). In the following sections I will not be primarly concerned with questions as to how particular metaphors function or what their general linguistic structure is like.[6] Rather, the concern of the present analysis will be to examine as exactly as possible the general relationship between metaphors and similes—since they represent the figurative trope most closely related to metaphors—and to open up a perspective for viewing the conditions necessary to an understanding of metaphor as the most fundamental of tropes. For, as I believe, only on the basis of such an understanding will it be possible to acknowledge the full semantic, aesthetic, and cognitive range of metaphors which so deeply concerns different researchers investigating the role of metaphors in philosophy and science.[7]


2. Passing under the "Rainbow": Literal Comparisons, the 'Like', and Metaphors

The conventional way of treating metaphors and similes proceeds by remarking that if we compare a metaphor to the corresponding simile we see that the former trope appears to be in some way stronger than the latter one. Thus in sentences like "He is a wolf" or, as Shakespeare's Romeo expresses himself, "Juliet is the sun" we intuitively grasp that these sentences are, in form and content, "stronger" or more expressive than "He is like a wolf" or "Juliet is like the sun". Typically, the word "stronger" means a difference in the strength of assertion, or, as Tirrell says, in the assertional commitment to what is said.[8]

However, the assertional commitment to what is said by metaphors and similes seems to depend on different types of predication and, hence, cannot be a matter of mere subjective commitment. Therefore, the gradual difference in the strength of assertion, if there is any, should also be understood as an effect of different types of assertion, though the psychological effect of assertions on hearers may be a matter of gradual difference in the sense that we consider (or feel) it as "stronger" to say "X is Y" than "X is like Y". Nevertheless it does not seem justified, on the basis of such a psychological effect, to take the difference in assertion types between "X is Y" and "X is like Y" for a difference in "the strength of commitment". When this is done, different linguistic levels are conflated, those of the grammatical forms of predication and their respective functions. To say "X is like Y" is not merely to use a "weaker" predication form than "X is Y", but to use a different type of assertion, which can or may, if compared to the predication form "X is Y", effectuate either a weaker assertion commitment to what is said by the speaker or a weaker effect on the audience. Therefore, instead of reducing the difference in grammatical forms of assertion (predication vs. comparison) into one—presumingly identical—linguistic function, we should assume that the differences between metaphors and similes are twofold: they are different both in grammatical type and, consequently, in assertional commitment. But the latter is secondary to and dependent on the difference in the grammatical type of assertion.[9]

In saying this, nothing more or new has been asserted of metaphors and similes than would not hold of literal assertions and comparison statements. We encounter the same difference in the type of predication in assertions such as the following ones:

1(a): He is a preacher.

1(b): He is like a preacher.

1(c): He looks like a preacher.

1(d): He talks like a preacher preaches.


1(n): (X + verb phrase + like a Y)

In dealing with assertions of this kind we recognize at once that, in spite of their literal appearance, they are ambiguous in several respects. First of all, the sample 1(a) can be interpreted both as a literal statement or as a metaphor, depending on whether the person of which it is said is really a preacher or not, and it would not contribute to the understanding of such a context-free sentence if we replaced the personal pronoun "he", an indexical term, with a proper name (e.g. "John") or with a referring expression (e.g. "this man", "this person"). But regardless of this ambiguity, which is context-dependent, another kind of ambiguity makes itself noticeable when we compare the samples listed above. While it is not clear whether the sample 1(a), when taken in isolation, is a literal assertion about one person being a preacher or, possibly, an example of figurative assertion about the same person which relates her to a preacher implying that she might not be, other samples containing the semantic marker 'like' state clearly—and independently of any context—that the person referred to by the pronoun "he" is not a preacher. As we shall see later on, in the figurative context the implications will be diametrically opposed.

Thus we are able to observe an intriguing effect of the semantic marker 'like' which is neither identical nor reducible to its linguistic function (comparison between two unrelated or different things). Namely, while in samples 1(b-d) the semantic marker 'like' disambiguates the context-dependent ambiguity in 1(a), it provides at the same time—by being a common element of 1(b-d)—a discursive context for 1(a) to be interpreted metaphorically. For we may inverse the order of the assertions which results in the order 1(d—>a) instead of 1(a—>d), so that, instead of the initial poly-univocity (i. e., sets of different, but univocal senses, either literal or metaphoric or even both ) we are now confronted with a kind of uni-equivocity: the inverted sample order has the effect of unambiguously establishing the metaphoric character of assertion 1(a), thus allowing for further grammatical transformations and, accordingly, for further assertion procedures such as the one indicated by 1(n) in the figure below:

1(d): He talks like a preacher preaches.

1(c): He looks like a preacher.

1(b): He is like a preacher.

1(a): He is a preacher.


1(n): This preacher (...)

But the intriguing and peculiar difference in the effect of the semantic marker 'like' on the sample 1(a), which contains the simple form of predication ("is /a/ Y"), will become more obvious if we examine the following samples of figurative speech:

2(a): Juliet is the sun.

2(b): Juliet is like the sun.

2(c): Juliet smiles as the sun shines.


2(n): X + verb phrase + as + Y + verb phrase

It is notorious that parallel relations exist between metaphors, similes, literal assertions and comparison statements: metaphors relate to literal statements by sharing the same kind of predication (cf. 2(a)=Juliet is the sun :: 1(a)=He is a preacher), and similes relate to comparison statements by sharing the explicit semantic marker of similarity (2(b)=Juliet is like the sun :: 1(b)=He is like a preacher). On the other hand, metaphors relate to similes on the ground that they are both figurative expressions, while comparisons relate to literal assertions on the ground that both are literal forms of assertion. This proportional analogy between tropes and literal forms of comparison and assertion can be explained in as many ways as there are respects in which the predication can be analyzed (identification, attribution, existence) and in which semantic markers of similarity can be listed (is like, like, as, as much as, as many as, similar to etc.).[10]

However, a third analogy relation between the four elements is not proportional. Contrary to comparisons, metaphors do not contain a semantic marker but are figurative; on the other hand, similes, contrary to literal assertions, require an explicit semantic marker and yet are figurative. Hence, there seems to be no proportional relation and no passage from similes to literal assertions, or vice versa and, equally, no passage from metaphors to literal comparisons, or vice versa. But there seems to be a peculiar—and quite different—aspect in which the function of the semantic marker 'like' could be analyzed, and this is precisely the one we have observed in the samples 1(d->a) which showed the condition required to establish a passage from literal comparisons to metaphors. It is an effect of the semantic marker itself which makes the literal assertion, such as 1(a), a figurative expression. Hence, the disproportional relation between literal comparisons and metaphors turns out to be a relation of consistency. For we have until now been able to observe that the semantic marker 'like' has had the peculiarity of disambiguating the context-dependence of literal expressions. Regardless of our knowledge about the person referred to by the indexical term "he" in sentence 1(a), it became—by introducing the semantic marker 'like' and thereby extending 1(a) to 1(b)—unambiguously clear that the person in 1(a) was not a preacher. But it also became unambiguously clear that, given 1(d—>b), the expression 1(a) had to be a sample of metaphoric speech or, at least, a trope-like assertion which loses its context-dependent ambiguity by exchanging it for a semantic one.

My suggestion so far is that assertions like 1(a)(=He is a preacher) are—unlike assertions such as 2(a)(=Juliet is the sun), where general knowledge or dictionary and encyclopedia entries provide sufficient conditions for understanding—literal assertions which depend on the context in which they are stated or applied to a person. Contrary to this, we have seen that by virtue of the discursive context which is provided by assertions 1(d—>b) with the semantic marker 'like', the assertion 1(a), supposedly literal, was revealed to be metaphoric. But at this point the question may arise whether it is sufficient for a literal expression such as "He is a preacher" to be interpreted metaphorically if we are sure that the content asserted (being a preacher) is not the case and that the assertion itself is not due to a false statement or a lie. In other words, the question is: What enables us to speak of the person referred to by "he" in such a way that literal assertions about this person 1(b—>d) allow for a passage to a wholly different form of predication (copula + noun phrase in 1(a)) and, moreover, for the introduction of a completely different referring expression (cf. the attributive term: "This preacher" in 1(n), instead of "He"). To put it more generally, what enables a literal expression to become metaphoric, to refer and to have a reliable meaning? An answer to this question may be approached by reconsidering the role of the "literal" semantic marker within a broader pragmatic framework and by relating our understanding of this role to expressions which may be considered as univocally metaphoric, such as 2(a)=Juliet is the sun.

The effect of the semantic marker 'like' in metaphoric utterences, if related to corresponding similes, seems not to be the same as in literal assertions when related to comparison statements. If we reconsider the relationship between metaphoric expressions and similes as in 2(a-c), then we see that there is no passage from comparison statements 1(d—>b) to the literal statement 1(a) unless the subject referred to, or the truth conditions of the sentence, change. This means: a passage from 1(d—>b) to 1(a), is not possible without a change of reference on the side of the subject referred to, or a change of knowledge conditions on the side of the speaker. Thus, if we say of a person that "He is like a preacher" we cannot pass over to the statement "He is a preacher", and mean it literally, unless we refer to another person or somebody else augments our knowledge of the former person by saying "But he is a preacher". Hence, we may assume that the semantic marker of similarity in literal comparison statements, while indicating that "is Y" is not the case, allows for the possiblility that "X is Y" be the case. But this possibility, to be real, necessarily requires a change of purely linguistic truth conditions into historic context conditions. If "is Y" is the case, then the literal sentences "is like Y" change their assertional status from literal comparisons into hyperbolic or emphatic expressions.

With this background we can assume that the semantic marker 'like', while disambiguating the context-dependent ambiguity of literal statements such as 1(a) (="He is a preacher"), and introducing a meaning related pluri-monosemy, which is represented by sentences 1(b-d), has the effect of preventing a straightforward passage from literal comparison statements to literal assertions, and vice versa, if the subject of reference remains identical. This means that the introduction of the semantic marker into literal discourse brings about a change of truth conditions, such that, while applying "X is Y" and "X is like Y" to the same subject, we either accept that the subject of reference must be different each time or we replace one type of predication with another and take into account that our assertional commitment cannot be the same in both cases ("knowing that" vs. "feeling that"). Only in this case can we, as I believe, comprehensively speak of a gradual difference in the assertional commitment to what is said.[11] But this assumption also touches upon literal statements and literal comparisons if it is expected that each of them relate to the same subject of reference. Thus, on the one hand, we say of a person that he/she is like a preacher only on the condition that we do not know about his/her really being a preacher; on the other hand, if we know that he/she is a preacher we normally do not use comparison statements containing the semantic marker of similarity to refer to the same person except in order to emphasize our assertion. Thus we may say: "He is (or: He may be) a preacher, but he also is (looks, behaves, talks etc.) like one". In that case, however, we extend the discourse by experimenting with its truth conditions, and, as a matter of fact, we intend to say something different of the person referred to: not only that he/she is a preacher, but that he/she is like a preacher, meaning that he/she is a prototype of the preacher. Although this latter possibility is semantically closely related to sentences like 1(a) and to comparison statements 1(b—>d), this meaning is nevertheless an effect of a different type of assertion, combining two gradually (cf. "not only, but", "moreover" etc.) ordered types of predication (assertion vs. comparison statement). But in such samples of hyperbolic stylisation of the asserting procedure, a possible extension of the discourse may by provided by a conversational reply: "A preacher is a preacher, so every preacher is like a preacher". In this way we weaken the assertional commitment of the hyperbolic (doubled) predication form "is and is like", by indicating that "is" entails "is like", such that being a preacher entails some similarity to the commonly assumed properties of being a preacher (be they related to appearance, behavior or psychology).

But at this point we realize that not every preacher-like person is a preacher. If it is not apparent that the referent is a preacher, we cannot, on the ground of literal comparison procedure (cf. "He is like a preacher"), make the literal and true statement "He is a preacher". Given these conditions we can only make an uncertain statement that may be true or may be false. Hence we can plausibly assume that the semantic marker 'like', if added to a literal statement, produces two complex effects: first, it disambiguates the contextual ambiguity of assertions but with the result of literalizing or, more precisely, providing literal truth conditions for comparison statements (in the sense that "being like a preacher" is equivalent to having some aspects in common with a preacher); second, it provides the necessary condition for the initial literal statements to be metaphorical. The condition consists in this: the semantic marker 'like', though relating to the same subject of reference as the simple predication form "is", disconnects the initial literal statement 'X is Y' from the ontological presupposition; namely, that the subject of reference actually is that which is predicated of the subject. In this respect we can say that there is no straightforward transition from literal comparisons to literal assertions. As we have seen, this effect of the semantic marker 'like' is context-independent: if we say of a person that she is like Y, we presuppose that she is not Y, irrespective of whether she actually is or possibly may be. And it is just this presupposition which provides the necessary precondition for the corresponding literal assertion 'X is Y' to be interpretable as a metaphoric expression.[12] Thus we may generally assume that, in a literal environment, the semantic marker of similarity will play the role of both literalizer—providing that comparison statements remain literal—and metaphorizer—providing that the place of the corresponding literal assertion is occupied by a metaphor candidate.


3. The Double Role of the Semantic Marker and the Likeness-Implication in Metaphors

If this is true, we have, in a metaphorical environment, quite a different situation with respect to the role of the semantic marker of similarity. While passing from the metaphoric expression 2(a) to the corresponding simile-versions 2(b—>c), there is no oddity comparable to the literal environment: in saying 2(a) "Juliet is the sun" we do not find anything unsound with respect either to the status of the subject of reference in statements such as 2(b) "Juliet is like the sun", or to the veritative status of attributes predicated of the subject, for "Juliet" remains the same in 2(a) and 2(b).

The above statement can be better approached by analyzing the relationship between the simple predication simile and corresponding similes which carry grammatical and lexical transformations of the predication form "is like" through other verb or noun phrases. To proceed, let us reconsider the simple simile form of the much exploited metaphor by Shakespeare cited above as 2(a) "Juliet is the sun". It is not an isolated metaphoric expression but part of a discursive context which allows for more precise interpretation of the metaphor than would be possible if it were an isolated sentence. In the corresponding verses of Shakespeare's masterpiece, Romeo says:

"Ah, what light through yonder window breaks,

It is the East, and Juliet is the sun".

Thus, before introducing the Juliet-metaphor, the poet first refers to the morning light breaking through the window of Juliet's room, calling it "the East". It is precisely the fact that metaphors are embedded in a framework containing other metaphors and related to other metaphors (and similes) in order to build a broader metaphorical discourse, that is decisive in accounting for the specific linguistic function of a particular metaphor. But, for the purpose of the present analysis, we can focus on the corresponding simple predicative simile which, as a matter of fact, Shakespeare did not use. If he had, then it would have been the one of the form

"Juliet is like the sun" [= 2(b)],

and we would have, among other hermeneutic tasks, to interpret the respect in which Shakespeare intended the likeness between Juliet and the sun. However, what is decisive from the standpoint of a formal analysis is not primarly whether we succeed in interpreting the non-Shakespearean simile by restating the similarity the poet might have had in mind or by listing all possible aspects which could account for such a simile irrespective of historical or cultural conditions. Rather, of far greater importance is whether we understand the grammatical form and the lingustic function of the simile, since the necessary—though not sufficient—precondition for interpreting the simile in as many different ways as possible is that we are able to maintain the same grammatical form and linguistic function of the semantic marker. Or, to put it more precisely, we are allowed to interpret a particular simile under the condition that we have understood its grammatical form and its linguistic function (which is, its peculiar role in our grasping "the world"). If we have done so, then we may—and we actually do—interpret the figurative speech sample independently of the respective author's intentions. And it is only due to this condition that we are able either to search for aspects of likeness which purport to be historically intended or to feel justified in neglecting the historical question as to what Shakespeare's (or any author's) subjective intentions might have been.[13]

Hence, the above-cited simile may be comprehensively supplemented by further versions preserving its grammatical form:

2(b) Juliet is like the sun.

2(b)-1: Juliet is warm like a sunny day.

2(b)-2: Juliet's face is brillant like the sun-shine (in the morning).

2(b)-3: Juliet's hair is like the sun's rays falling etc.


2(b)-n: (Juliet's + np  + is like + np + etc.)

What is striking in the examples listed above is that the subject of reference, Juliet, is given literal properties (warm, shining) or literally attributed substantives (physical world items: rays, sunny day) of the sun but not figurative ones.[14] Thus we see that what makes the properties of the sun figurative is only their being attributed to Juliet who is a young female human being and not an asteroid. In this respect one can say that the simile expressing a figuratively intended comparison—this means: embedded in a metaphorical framework of predication (Juliet vs. the sun)—relies on literally comprehended properties or attributes of the sun. In saying this we should not be concerned with questions as to whether comparison and likeness statements really imply all literally predicated properties of the Y-element in the simile: they clearly do not, for to say of Juliet that she "is (like) the sun", does not and cannot, in the given discursive context (a tragedy), imply in any way that she is imagined as an immense burning asteroid or that she is positioned far away in outer space. Admittedly, such implications are not precluded both from the perspective of the speaker and of the listener, but if they were to appear in the given discursive context, they would necessarily cause a change in character of the discourse itself (e.g., transforming the tragedy in part or as a whole to another genre).[15] As we know, such subversive implications are allowed and even integrated in comedies, parodies and in speech samples based on everyday language use. This means that every competent speaker of a language can make such subversive implications and build a correspondingly subversive (or deconstructivist) discourse. But what is important in such procedures is that we, by subverting the discourse, wittingly prove that we have understood the figurative speech sample: for we know which implications of a figurative expression are precluded by the given context just as we know the same for the literal environment. Thus it is just this fact which justifies the assumption that—although we are not obliged to take into account all actual properties of an Y-element of the predication or all possible ones—all the properties of Y we actually do apply to the X-element of the figurative expression are literal. The precondition for this is, of course, that all properties at issue are properties of or properties attributed to Y, for it is in no way precluded that X be given properties of Y that are figuratively attributed to it (indicated below as *F). Hence, we may say:

2(b)-n1: Juliet is like the smiling sun.

2(b)-n2: Juliet's burning arms are like the sun's rays eager to embrace the earth.


2(b)-nn: (Juliet's + np*F + is like + np*F + vp*F etc.)

These samples do not contradict the assumption that in similes the predication procedure relies upon applying literal properties of Y to X. They are more complex similes, built upon other tropes which can be resolved by analyzing the properties figuratively applied either to Y ("the sun's rays eager to embrace") or to X ("burning arms"), if there are any. But what is intriguing in all these examples is not whether the particular properties of Y, applied to X, are literal or figurative. For, whether literal or figurative, the intrinsic property of the simile as a figurative comparison is retained because of the character of the terms related ('Juliet' vs. 'the sun'). Moreover, it is clear that figurative properties of Y may be taken from X itself (as in 2(b)-1: "smiling sun"), from other contexts such as mythology, or from other objects in the world.

Hence, what we may assume as strikingly true of similes is that it is not primarily the figurative character of properties of the Y-element that constitute their figurative status; much less trivial of similes is that they, as examples of comparison operating on properties of unrelated things, accomplish, by their very grammatical form, a literalization of the predication procedure which indicates a particular similarity relation between X and Y. Or, to put it more clearly, it is not the property of the sun as being a burning (hence: hot and bright, warm and shining) asteroid which is equated with any intrinsic property of Juliet as a female and a young human being. It is exactly the semantic marker installing a similarity relation between two discrete items in the world which generates the equation. Whatever the properties of Y are like in character, their linguistic application to X is performed and governed by the semantic marker of similarity. In other words, the only visible linguistic factor to define the character of the expression and to enable that unrelated things like Juliet and the sun be related to each other is the comparison marker itself, which represents, of course, the linguistic function of comparison intended by the speaker. But, as we know, the sematnic marker of "comparison" is capable of connecting and relating quite unrelated things. Hence in reading "X is like Y" we can only read that "X is like Y" and not that "X is Y". But the precondition for us to grasp that one particular expression establishes a comparison or defines a similarity relation between X and Y, and not an identification or any other linguistic function, is to grasp the grammatical form of the expression.[16]

But, as we know, this grammatical form is common to simile as well as to literal comparisons, which implies that it is not specific to similes and not by itself able to reveal a difference between similes and comparisons (cf. Tirrell 1991). This finding is nevertheless everything but trivial. We have enough reason to believe that the intriguing question about similes is not whether the properties attributed to the X-element are literal or figurative properties of the Y-element, but rather what the effect of relating X and Y via the semantic marker of similarity is. In the same way that the literal environment, presented above, was examined with the help of comparison statements 1(b-d), we shall look at the function of the semantic marker in the simile with respect to the corresponding metaphor. For we know, at least by intuition, that relating the properties of one item of the world to another does not by itself cause any striking difficulty in our understanding of the world. This is due to the fact that our conceptual apparatus itself is built upon metaphoric patterns so that we relate to abstract objects in the way we relate to physical objects: thus we "come" to ideas and we "approach" problems.[17] In this respect figurative similes and genuine metaphoric expressions do not essentially differ. This continual passage from "related" to "distantly related" and "not related" things is not, as I have suggested, due to the properties of Y which are attributed to X, but seems to be an effect of the peculiar function of the semantic marker.

The statement that a girl "is the sun" does not convey anything unsound with respect to the statement 2(b) "Juliet is like the sun", provided that we have contextual knowledge enabling us to understand and process the personal pronouns and proper names into referring terms, and vice versa. More precisely, what we observe while passing from metaphors to similes, is that the sameness of both the referent and the attribute predicated of it, despite the transition in tropes, results from the fact that here the semantic marker 'like' has no bearing on the ontological status of the referent's being Y. Juliet's "being the sun" and her "being like the sun" are, according to our present knowledge of the world, equally impossible. Accordingly, we cannot say that it is the semantic marker 'like' in similes 2(b-c) which by itself makes the expression 2(a) metaphorical, as was the case in literal comparisons such as "He is like a preacher". Nonetheless, the similes are, unlike metaphors, essentially constituted by the presence of an explicit semantic marker, just like comparative statements, and what we observe now is that the semantic marker 'like' turns out to be wholly irrelevant for an expression to be considered figurative, whether it is a simile or a metaphor. Hence, the genuinely metaphoric status of an expression such as 2(a)=Juliet is the sun, unlike the ambiguously literal one (1(a)=He is a preacher), appears wholly independent of any possible relationship to a corresponding simile, although the related simile may, or actually does, contain the same elements as the corresponding metaphor, such that the metaphor 2(a) "Juliet is the sun" has the appearance of being identical with the simile 2(b) "Juliet is like the sun", except that it contains an additional grammatical element.

On this basis we can conclude that since similes are necessarily but not exclusively constituted by the presence of an explicit semantic marker, they are formally related to comparative statements and not to metaphors. If we remember that metaphoric expressions behave grammatically in the same way as literal statements do—they contain the same pattern of predication and allow for the same transformation of predicates into attributes[18]—we see that the proportional analogy between tropes, mentioned above, holds also in another respect, namely that metaphors relate to literal assertions in the same way similes relate to comparison statements, a reason for this being that they belong to different modes of predication.[19] Thus, while it is obvious that similes necessarily contain an explicit semantic marker of similarity just as comparison statements do, one must pay full attention to the fact that corresponding metaphors share the same grammatical form with literal assertions and not with figurative similes. But by occupying opposite places in the positions layed down by the analogy symbol (::), there arise further serious consequences for the conventional wisdom concerning the relation of metaphor to simile.

4. Beyond the 'Like': Metaphors unlike Similes

What has been established thus far is that it is not constitutive of metaphors to be reducible to similes but, instead, to share the same grammatical forms of assertion with literal ones. This, however, does not imply that metaphors and similes are wholly unrelated. Rather it suggests that their relationship is different than has usually been assumed. So far the ground has been prepared for the notion that the semantic marker of similarity, although constitutive of similes, is not only inessential to the figurative character of tropes, but—being essential to the linguistic function of comparison—grammatically and semantically at odds with the predication form of metaphors. This implies that, at least in some contexts, the use of the semantic marker might block the formation of metaphors, and vice versa.

In order to approach these aspects of metaphor we may state that, in contrast to the literal assertion 1(a)=He is a preacher, samples of the simile 2(b-c) provide no discursive context for the corresponding metaphor 2(a)=Juliet is the sun. In other words, a statement such as "Juliet is like the sun" contributes nothing either to the referential, or veridical status, or to the understandability of the metaphorical statement "Juliet is the sun". If it did, then we would have to presuppose that the former sentence is a more comprehensive version of the latter one. But we are not told in which respect Juliet is like the sun, and we surely do not feel much more comfortable with the content of this statement than with the content of the metaphor itself. But even if it were the case that simple similes are more comprehensible than metaphors—and it is clearly not, as Fogelin (1994) and Indurkhya (1992) correctly assume—the only reason for this greater comprehensiblity of the simple (and abstract!) simile would be our understanding of the linguistic function of the semantic marker itself, for it constitutes the only overt difference between similes and metaphors.[20]

But despite such a close relation between metaphors and similes, a relation which tempts us to assume that similes are metaphors plus the 'like', we hold that, though we are not told in what respect X is like Y, we are—by the very grammatical form of the simile—given discourse-related information, namely that X and Y are related to one another by virtue of the assumption that they are alike. This obviates the assumption, told us by the corresponding metaphor, that they are identical. Thus we have several important elements for further analysis: first, what is informative or content-providing in both the metaphor and the corresponding simile is only their respective grammatical form; this form is either that of predication (X is Y), on the one hand, or that of a similarity statement (X is like Y), on the other. Second, what the similarity marker aims at is not to indicate the particular is-relation between X and Y (be it similarity, identity, partial identity), but rather to specify the aspects of the similarity relation itself, i. e., aspects of likeness, that do pertain between X and Y in the given simile. Hence, third, the similarity marker in a simple predicative simile establishes and guarantees only the similarity relation itself and requires that instantiations of it be displayed by pointing out particular properties of Y. Thus, even when confronted with a radically unusual simile, we either accept the "likeness" intuitively or we searche for aspects of "likeness" between unrelated things, i. e., we reflect upon the likeness suggested by the simile. On the basis of this, fourth, the simple predicative or metaphor-like simile, purporting to rely upon a simple extension of the predicative metaphor ("is like" for "is"), is nonetheless clearly directed towards other related (or derived) examples of simile and not to the "corresponding" metaphor. In other words, what the predicate phrase "is like Y" in a simile seems to suggest is either that "is" ought to be replaced with a more appropriate verb phrase or that Y should be replaced with a more appropriate noun phrase. The semantic marker in a simple predicative simile requires, for its comprehension, pointing out aspects of likeness such as the ones given in 2(b-c). Thus, fifth, since the copula in "is like" stands for another finite verb to describe one particular aspect of similarity between X and Y ('being like' instead of 'looking like', 'walking like', 'eating like'), it seems possible to assume that the simple predicative simile, containing only the copula 'is' and thus conveying a quite abstract or unspecified content, appears to be the reduction or ellipsis of corresponding extended similes.[21] But this is precisely what cannot be said of the "corresponding" metaphor because the copula 'is' is not an ellipsis of 'is like' for the simple reason that it performs a different linguistic function than comparison, and 'is like' is, by virtue of its linguistic function, related to corresponding semantic markers of similarity such as "similar to", "as" etc. Contrary to this, in metaphors things are not compared but related to one another by predication, identified with or subordinated to one another, and replaced or represented by one another.

It thus becomes obvious that the gap between similes and metaphors cannot be compared to the gap existing between a comparison statement and the corresponding literal assertion. The passage from a simile to the corresponding metaphor in no way changes the referential status of the subject or the veridical value of the statement itself. What has been demonstrated is rather the opposite, namely, that metaphors provide a discursive context for similes such that the semantic marker of similarity, if added to a genuine metaphor, causes, first, a change in linguistic function between metaphor and simile (predication vs. comparison) and thereby, second, a change within the figurative status of the comparison trope: while metaphors live from oddity ("figurativeness"), similes—assumed to be figurative—reveal themselves as examples of a literally intended linguistic function of comparison irrespective of the figurative status of properties of Y attributed to X. For, as we have seen in 2(b-c) and 2(b):1-n, it is not that properties of the sun—be they literal, figurative, or combined—are taken as properties or "real attributes" of Juliet, but that they are applied to Juliet via the similarity marker. In other words, the only element taken as literal is the linguistic function of the semantic marker itself, i. e. the comparison. This entails—apparently contrary to our assumptions so far—that, third, if the simple is-predication in metaphoric expressions is given a similarity marker, a change in the ontological status of the referent X in the sense of "being possibly the case" must be the the effect: namely, while in metaphors we know that 'X is Y' is not the case, in similes we accept or deny the possibility that X may be like Y. This means that metaphors and similes entail respectively different ontological presuppositions ('is not Y' vs. 'is like Y possibly') and, consequently, that their relationship should be redescribed by different systematic means than the similarity assumption.

In this respect the semantic marker used in similes plays just the opposite role that it does in the literal environment: it transposes the asserted figurative is-relation between X and Y, processed exclusively by the predicative linguistic function of metaphor, into the literal function of comparison. Therefore it is possible to state that metaphors are antecendent to similes because they provide the necessary onto-logical condition for figuratively ascribing likeness or similarity to unlike things. We know that Juliet is not the sun, and this knowledge is what makes metaphors possible as well as generally comprehensible. We accept that she, though being totally different, may be like the sun, just because a certain relation between Juliet and the sun has already been processed through the metaphor. Hence, the only element capable of "comparing" the properties of unrelated things is the semantic marker itself, which keeps a place open for selecting ever new aspects of "likeness". It gives the necessary linguistic framework for "untenable" comparisons, but it neither constitutes the metaphor, nor compels its comprehension, since no like-relation or property can itself be the bridge for passing on to an is-predication of the corresponding metaphor. This holds true in the literal environment because, as we have seen, the like-relation in "X is like Y" indicates that "X is (a) Y" is not or possibly not the case. In the metaphoric environment the same holds but for quite opposite reasons: it is the presupposition of not being the case, introduced by the corresponding metaphor, which permits similarity relation to be asserted at all, i. e., which permits that the impossible is-relation of the metaphor becomes the possibly true like-relation of the simile.

5. Metaphors : Filling the Gap with Chasm

The subverted relation between metaphors and similes, as indicated above, becomes more transparent if we remember that similes such as 2(b-c) and 2(b):1-n, if read in the reverse order, do not bring about the linguistic status of the expression 2(a), as was true of 1(d—>a). If similes were the way to pass on to metaphors, it would at some point be possible for the metaphoric expression 2(a)="Juliet is the sun" to turn out to be a literal expression. For, if some day enough intrinsic properties of the sun could be attributed to Juliet, it would arouse our suspicion about the referential relation of the proper name "Juliet", and make us wonder whether it might not actually designate an asteroid rather than a young female human being. But, as has become clear, exactly the opposite is the case: it is the conjunction of two unrelated items of the world, themselves conjoined only by means of the grammatical form of predication, that governs the attribution of "properties" of the one to the other. If Juliet is like the sun, we assume—just as in the literal context—that she is not the sun and that she may be the sun only under the condition that "Juliet" refers to an asteroid. In this case the similes would change the referential status of the expression, assumed to be metaphoric, into a literal one.

But this is impossible by the very fact that—contrary to literal environments such as 1(a-d) where it is not ruled out that "he" may indeed be a preacher—a young female human being cannot be an asteroid. As a consequence, we cannot explain the passage from the predicative simile "X is like Y" to the predicative metaphor "X is Y", because in proceeding from similes to the "corresponding" metaphor we only better understand and describe the "similarity" assumed to hold between X and Y, but we do not arrive at the metaphor itself. In other words, additional (more concrete and more comprehensible) similes only represent and explain—under different and more concrete aspects—the similarity relation which is stated by the simple (and abstract) predicative simile, but they do not allow for a passage to the metaphor itself. Rather it is the metaphor which governs and limits the domain of permissible properties of likeness by simply imposing a relation between X-elements and Y-elements.

On the basis of this discussion we may restate that, within the figurative environment, the formation of discursive comprehensibility conditions proceeds in the opposite direction when compared with the literal environment. This means that similes do not provide a discursive context for understanding the status of the corresponding metaphor as comparison statements do for the corresponding literal statement. While extending the similarity aspects through more concrete similes we only attain a better, more elaborate understanding of the similarity relation, but we do not pass from "is like" to "is", i. e., we do not translate the linguistic function of simile into the linguistic function of metaphor. Therefore there is no reason to believe that the linguistic status of metaphor depends in any way on how well or poorly we may understand a simile. If the linguistic status did depend on such an understanding, then metaphor would be either a candidate for the ascription of truth or falsity,[22] or it would require another form of discourse to become plausible (such as tales and myths), or other worlds for it to become possible.[23] Hence we recognize that the precondition which allows the creation and the comprehension of metaphors, in spite of their semantic oddity, is the ontological presupposition of their not being the case. The means a metaphor uses to show up this impossibility is nothing other than the linguistic function of predication, which it shares with the literal discourse. Accordingly, the comprehensibility which relies on the possibility that a likeness-relation pertains between X and Y is not necessary for a metaphor to be a metaphor. It is, quite to the contrary, the "untenable" form or the linguistic framework which is necessary—though not sufficient—for metaphors to be linguistically possible and comprehensible. This entails that metaphors are dependent on other truth conditions (if there are any for metaphors) and on other linguistic function (if there is a particular "metaphorical" one).

In assuming this, however, nothing has been said about the ultimate status of the similarity implication in metaphors. We have only implied that similarity must relate to the comprehensibility and the acceptability of metaphors in a different way than it does in the case of similes. The usual explanation of this problem, characteristic of nonreductive simile theories of metaphor, is to say that, while similes contain an explicit marker of likeness, metaphors rely upon the implication of likeness, whereby it is not necessary to assume that likeness or similarity can explain the whole meaning of the metaphor; the implication of likeness is considered to provide only the necessary cognitive background ("semantic memory" or "knowledge of world") for understanding metaphors, since metaphors impose, like all new information and knowledge, apperceptive problems on the cognitive process of human beings (cf. Miller 1979). If related to our account as presented so far, this explanation seems to say that instead of being a constitutive element of the grammatical structure as in the case of similes, similarity delimits the horizon in which cognitive judgements and psychological expectations concerning the meaning of metaphor are formed. Thus, the similarity assumption, by exchanging the presence—given by the very grammatical structure of the simile—for absence—the condition of all metaphors—transforms the ontological status of metaphor from "non-being" to "being". This transformation is the consequence of the linguistic function performed by the particle 'like'. By thus founding similes as well as metaphors on the similarity implication, all comparativist accounts of metaphor, whether reductive or not, must make use of the operator of existence (cf. Miller).

But this is fundamentally misleading. Not because metaphors are absolutely unrelated to the similarity implication or to cognitive processes, both of which are based on apperceptive processes; but only because the ontological presupposition of "not being the case" is the precondition of producing, understanding, and accepting metaphors just as the ontological presupposition of "being the case" is the precondition of forming and accepting literal assertions. If we remember that there are, as a matter of fact, no similarities between an asteroid and a female human being to be common properties of the sun and Juliet—for if there were even one single material similarity or analogy, the simile would necessarily turn out to be a literal comparison—we will see that the similarity or likeness-relation between the terms related is not processed for metaphors through similes. Instead, we may state that the similarity relation between the terms is imposed onto similes through the metaphor, which means that metaphors govern similes by defining the limits of likeness, and not vice versa. Metaphors precede similes by being their onto-logical presupposition, and this is the reason why we can say that metaphors make room for the creation of similarity aspects rather than merely being derivative of already existent similarities, be they obvious or hidden. But this is also the reason why similes and metaphors, without their mutual reducibility being a necessary and true implication, are related to one another. These two reasons provide, in my opinion, the necessary, but missing, fundament for interactionist theories of metaphor (Kittay 1987, 1994 and Indurkhya 1992, 1994) which insist that metaphors, at least novel ones, are essentially characterized by their ability to create new similarities, rather than by their dependence on already existing ones.[24]

Once we see this priority of metaphor as condition of possiblity of likeness-relations, we can understand that, although similes are also, at least in part, parasitic on the ontological precondition of "not being the case", this precondition is provided only by metaphors and processed into further linguistic functions (predication, attribution, reference) without relating to similes. We can also understand that it is this feature of metaphors which prevents similes, despite their sharing in part the same ontological presupposition with metaphors, from replacing or approximating metaphors in every discursive context or at any level of the same discourse. For although we may accept as true that there is not much difference in the comprehensibility of sentences such as "Juliet is the sun" and "Juliet is like the sun", we know that the grammatical function of "being like" cannot allow for the noun phrase "the sun" of the predicate to become an attributive and referring term capable of replacing "Juliet". The condition for this replacement can be provided only by the metaphor "Juliet is the sun", by virtue of its grammatical form. Nor can, for the same reason, Plato's figure of the sun in the Republic permit "the sun" to become the referring term of the highest principle. In order to make it possible, a metaphor must be in operation, and not a simile.

Hence, the tropological quandaries involved in the relationship between metaphors and similes have revealed themselves as profound family disturbances within related tropes: it is not similes which explain the metaphor, but it is metaphors that prepare the ground for similes to be linguistically permissible and intersubjectively comprehensible. The supposedly inexplicable difference in expressive strength between metaphors and similes, which caused so much trouble to friends of metaphor, appears to be the product of nothing more grandiose than the difference in linguistic function in predication and comparison. This difference indicates a deeper and more basic difference in the logical relationship between metaphors and similes, rendering metaphors antecendent to all figurative language use. In this sense, it is not constitutive of metaphors to relate to similes as a more extended or more "comprehensible" form of figurative speech—for, as we have seen, the only element of a simple predicative simile which is more comprehensible when compared with the corresponding metaphor is the semantic marker itself—but that metaphors are due to linguistic procedures which are characteristic of literal language use. Metaphors do open the linguistic framework and conditions of comprehensibility for similes. However, they operate at the same level—by their very linguistic function—as literal predication procedures but not as similes. Hence, similes reveal themselves to be "a false currency" in the attempt to explain the origin of metaphors as well as their comprehensibility and cognitive impact on our language and knowledge. Though metaphors themselves need not be considered unrelated to the similarity implication in similes, the similarity processing by similes does not explain why and how metaphors arise in language. This is the reason why passing through the 'like' on to metaphors only produces further similes but not metaphors. A metamorphosis of comparison tropes into metaphors is only possible by starting from the literal processing of likeness. Hence similes may allude to metaphors, but analyzing them on the basis of family resemblances between the tropes results in illusions about metaphors. The metaphoric function of language, if there is one, must be different from the linguistic function of similes.



(containing the literature mentioned or commented upon)

Black, M. (1962), Metaphor, in: Models and Metaphors, Ithaca UP; cf. also How Metaphors Work in: Sachs, S. (ed.), On Metaphor, Chicago: Chicago UP 1978; and More on Metaphor in: Ortony (ed.) 1979

Boyd, R. (1979), Metaphor and Theory Change: What is "Metaphor" a Metaphor for?, in: Ortony (ed.) 1979

Davidson, D. (1984), What Metaphors Mean in: Inquiries into Truth and Interpretation, Oxford UP

de Man, P. (1978), The Epistemology of Metaphor, in: Critical Inquiry 5 (1978)

Debatin, B. (1995), Die Rationalität der Metapher. Eine sprachphilosophische und kommunikationstheoretische Untersuchung, Berlin: de Gruyter

Derrida, J. (1972), La mythologie blanche, in: Marges de la Philosophie, Paris; also Le retrait de la métaphore in: Psyche. L'invention de l'autre Paris 1987

Fogelin, R. (1994), Metaphors, Similes, and Similarity, in: Hintikka (ed.) 1994

Haack, S. (1994), "Dry Truth and Real Knowledge". Epistemologies of Metaphor and Metaphors of Epistemology, in: J. Hintikka (ed.) 1994

Hintikka J./Sandu, G. (1994), Metaphor and Other Kinds of Nonliteral Meaning, in: Hintikka (ed.) 1994

Hintikka, J. (1994) (ed.), Aspects of Metaphor, Dordrecht/Boston/London: Kluwer Academic Publishers

Indurkhya, B. (1993), Metaphor and Cognition, Amsterdam 1993

Indurkhya, B. (1994), Metaphor as Change of Representation: An Interaction Theory of Cognition and Metaphor, in: Hintikka (ed.) 1994

Kittay, E. F. (1987), Metaphor. Its Cognitive Force and Linguistic Structure, Oxford University Press

Kittay, E. F. /Steinhardt, E. (1994), Generating Metaphors from Networks. A Formal Interpretation of the Semantic Field Theory of Metaphor, in: Hintikka (ed.) 1994

Lacks, A. (1994), Substitution et connaissance: Une interprétation unitaire (ou presque) de la théorie aristotélicienne de la métaphore, in: Furley/Nehamas (eds.), Aristotle's Rhetoric. Philosophical Essays, Princeton 1994

Lakoff, G. (1995), How Conservatives Left Democrats in Dust, in: "The Power of Metaphor" in: Social Research Vol. 62, No 2 (Summer 1995)

Lakoff, G./Johnson, M. (1981), Metaphors We Live By, Chicago and London: The Chicago UP

Lloyd, G. E. R. (1987), Metaphor and the language of science, in: The Revolutions of Wisdom. Studies in the Claims and Practice of Ancient Greek Science, Berkeley: Univ. of California Press

Mikulic, B. (1995), Politics of Metaphor: Providing Political Power by Re-Literalizing Metaphors in Public Speech, in: Inquiry in Progress, ed. by J. Weinstein, Institute for Human Sciences, Vienna

Miller, G. A. (1979), Images and Models, Similes and Metaphors, in: Ortony, A. (ed.) 1979

Ortony, A. (ed.) (1979), Metaphor and Thought, Cambridge: Cambridge UP 1979 (repr. 1986)

Ortony, A. et al. (1978), Metaphor. Theoretical and Empirical research, in: Psychological Bulletin, 85

Ortony, A. (ed) (1979), Metaphor and Thought, Cambridge Univ. Press

Radman, Z. (ed) 1995, From the Metaphorical Point of View. A Multidisciplinary Approach to the Cognitive Content of Metaphor, Berlin: de Gruyter

Schön, D. A. (1979), Generative Metaphor: A Perspective on Problem-Setting in Social Policy, in: Ortony (ed.) (1979),

Searle, J. (1978), Metaphor, in: Erkenntnis 13 (1978), and in: Ortony (ed.) 1979

Social Research Vol. 62, No 2 (Summer 1995): "The Power of Metaphor"

Tirrell, L. (1991), Reductive and Nonreductive Simile Theories of Metaphor, in: The Journal of Philosophy No. 7, 1991

Winner, E. et al. (1976), The development of metaphoric understanding, in: Developmental Psychology, 12; Winner, E. et. al. (1980), Misunderstanding Metaphor. What's the Problem? in: Journal of Experimental Child Psychology 30; and Winner, E. (1985), Metaphor and Irony, Oxford Univ. Press 1985.



|AppliedPhilosophy| |Courses| |Bibliography| |Biography| |Papers|


[1] Cf. Miller (1979), Fogelin (1994).

[2] Cf. Tirrell (1991).

[3] Cf. Davidson (1984). Although his position has been contested in central respects (cf. Kittay 1987) with arguments which seem quite convincing, I will adopt in this paper one of Davidson's claims about metaphors which insists that similes and metaphors are not reducible to one another, although he himself does not give sufficient reasons for this claim.

[4] In this respect, the implications (a)-(c) still conform not only to nonreductive similarity accounts of metaphor (Miller 1979, Fogelin 1994), but also to those theories of metaphor which claim that creating and interpreting metaphors require the same cognitive process as creating and interpreting figurative similes do (Kittay 1987, Tirrell 1991, Indurkhya 1992).

[5] On Aristotle cf. Lloyd (1987), and recently Lacks (1994); Haack (1994) on Locke and Hume; Paul de Man (1983) on Locke, Condillac, Kant, etc.

[6] For this, cf. the most comprehensive accounts Lakoff/Johnson (1981), Kittay (1987), and Indurkhya (1994). As for effects of metaphors on the social sphere cf. Schön (1979), Lakoff (1995); for the role of metaphor in building up a particular political discourse by means of impersonal interpellation cf. Mikulic (1995).

[7] For the most recent contributions cf. the essay collections "The Power of Metaphor" in: Social Research Vol. 62, No 2 (Summer 1995); further Radman (ed.) (1995) and Debatin (1995).

[8] Cf. Tirrell (1991). This assumption of difference in "assertional commitment" means that we must deal with a gradual difference in strength of commitment and not in the type of assertion, and that there is a continuity between similes and metaphors. But this seems inconsistent with Tirrell's explicit claim that we should assume a gap between, first, metaphors and comparisons, because metaphors do not entail any comparison statement, and, second, between metaphors and figurative similes, because, as she correctly assumes, the difference in semantic marker is not trivial, accounting as it does for divergent implications as well as for varying inference possibilities about the topic (or target).

[9] Tirrell (1991) operates with two kinds of difference in strength, a lingustic and a pragmatic one. Concerning the first, she states that "on a literal interpretation the 'like' weakens the claim to which it is added" (p. 352), while on the topic of the second she explains that "when interpreting a simple unextended simile the use of 'like' suggests a more limited endorsement" on the side of the audience such that "the audience [unlike the speaker], cannot tell which extensions are unavailable" (p. 354). Though these distinctions are illuminating, we shall see that the effect of the 'like' is not the weakening the assertional commitment but that of changing the type of assertion.

[10] Cf. metaphor : simile :: statement : comparison. This symmetric relation between tropes is also extendable to the grammatical functions of predication and similarity markers, such that it may include the lingusitic and logical relations: predication : similarity :: metaphor : simile :: statement : comparison.

[11] Again, it becomes clear that, when speaking of difference in assertional commitment, we assume an identity of lingusitic function between "is" and "is like". But this is clearly not the case for, first, in saying "is like Y" we do not purport to say "is Y", but state a similarity between X and Y, and this is a different kind of predication and not just a weaker commitment to assertion; second, it seems that "is like" is comparable or equal to "is" with respect to the assertional commitment. We observe this when somebody asserts that two persons are physically alike, and somebody else denies this. The assertional commitment of the speaker to his similarity statement about two other persons is not weaker than it would be in an "is" form of predication, although she never would say of the two persons compared that the one is the other. The reason is that what is literally meant is a similarity relation between different persons, and the similarity statement is meant to be literal. Hence, the "is like"-form of predication is quite different from the is-form, but equal in assertional commitment. In using one or the other predication form we actually intend to say different things (i. e., to apply different properties of X: being something and being like something). We struggle with is-predication as well as with is-like-statements.

[12] In order to indicate unambiguously that expressions such as "X is a Y" is a metaphor we usually and spontaneously use some other form of reference, such as the one indicated in example 3(n)="This preacher", where the predication form is replaced with the attributive position of the word. But whether we can make this transformation within a sentence or not depends on discursive context conditions or, more precisely, on whether the conditions are given for identifying the subject of reference by means of anaphora: e. g. This man … John … He … This preacher etc.

[13] But whether we do so or not, depends on the character of our analysis and of the so-called cognitive interest which may be historical, but need not be. This issue may be of interest in another context of metaphor analysis. For the moment it might be clear, however, that we have good reason to disagree with such accounts of metaphor as Searle's (1978), Davidson's (1984) or Haack's (1994), all of which coincide in assuming that the subjective intentions of the speaker or the hearer, or both, are decisive not only for creating but also for interpreting metaphors. The latter is, at least as a general claim for metaphors, not true. Contents of metaphors and figurative similes are partly dependent on subjective intentions, but the comprehensibility of these contents relies on the grammatical form, and not on intentions. The subjectivist theory of metaphor cannot account for the asymmetry between creating, but misunderstanding, as well as for poorly creating, but understanding, metaphors. Even less can it account for the fact that we understand and reconstruct metaphors in old texts although we are not able to reconstruct the subjective intentions of the respective writer. This fact indicates clearly that metaphors depend more on linguistic functions than subjective intentions, communication rules, and conversational maxims. Subjective intentions are not eliminable from metaphors, but they are governed by linguistic rules. For this cf. Kittay (1987).

[14] Tirrell (1991) wrongly takes single expressions like "brilliant" as a metaphor. This is not correct because we use the epithets "brilliant", "golden" etc. of jewels and other physical objects as well as of persons and abstract objects such as knowledge, books and states. Expressions such as "golden book", "golden goal" or "golden state" are not by themselves metaphors but idiomatic expressions.

[15] Hence, although we may agree with Tirrell that, in the case of the simple unextended simile, the interpretative position of the hearer is "weaker" than that of the speaker, we recognize that the hearer has the advantage of being able to intentionally misinterpret (and subvert) the speaker's own intention and thus to change the character and type of the discourse itself. In any case, the feature of weakness is not due to the linguistic function of the semantic marker 'like' (it expresses a different property of X than the 'is') but to the pragmatic conditions of the discourse.

[16] At this point the problem arises of the relation between the semantic field structure of our language and the role of linguistic functions within language or, subjectively speaking, between the so-called semantic memory and linguistic competence on the side of the speaker. So, if Kittay (1987) —in accordance with Chomsky but against Grice and Searle—assumes that the subjective intention of the speaker is not constitutive for the understanding of metaphoric expressions he uses, but that these expressions are due the his linguistic competence, it seems also necessary to assume, at the general linguistic level, another non-subjectivist consequence, namely the primacy of linguistic competence over the semantic memory given by the word field structure of a language. For, what in a language makes the choice of a particular word or an interpretation of the given sentence on the side of the listener possible, is the choice, by a speaker, of a particular linguistic function such as expressing, comparing, identifying etc. Thus it seems primarily the lingustic function, given by the respective grammatical form, which constitutes the linguistic identity of the given piece of language and which indicates (but does not absolutely define) the direction of interpretation.

[17] Cf. Lakoff/Johnson (1981).

[18] Cf. 'He is a preacher', 'He is a wolf'; 'The picture is blue', 'The picture is sad' -> 'This preacher', 'This wolf'; 'The blue picture', 'The sad picture' etc. Cf. also the so-called "dead metaphors" in idiomatic expressions: 'He spoke fluently', 'His speach was fluent', 'His fluent speech', etc.

[19] Cf. metaphor : statement :: simile : comparison

[20] Cf. the psychological studies by H. Winner (1976), and the diametrically opposed results by Ortony (1978) concerning the understanding children have of figurative comparisons in similes and metaphors. But if there is a significant asymmetry in a child's understanding of the two tropes, it might be due to the linguistic function of 'like' which provides that X and Y remain different things, while metaphors—equating X and Y—are counterfactual. Understanding metaphors would then require a higher linguistic competence which is a not only cognitive, since children seem capable of forming "metaphors" while believing that these are literal descriptions. Lacking the requisite ontological presupposition, these descriptions become children's "tales" and "lies".

[21] This is possible because we can form elliptic sentences with similes as well as with literal sentences. Thus, just as we scream "Fire!" instead of saying "The house is burning!", we also may form an ellipis such as "Like a preacher!" instead of the full comparison statement "He talks like a preacher!". Accordingly, we also may use a figurative ellipsis such as "Like the sun!" instead of e. g. "Her face is shining to me through the window like the sun rises in the morning!". The comprehensibility of all elliptic samples depends however on whether their reference is contextually defined.

[22] Davidson's (1984) thesis that metaphors, being a matter of language use and not of meaning, belong to the class of lies is simply wrong, because the pragmatic conditions for metaphors and lies are quite different. While metaphor entails the ontological presupposition of not-being, which must be obvious and accessible for both the speaker and the audience, a lie presupposes that "X is Y" may be the case. Examples such as "X is a communist" require, in order to determine whether they constitute lies or metaphors, much clearer context conditions and subjective beliefs than trivial metaphors. Nonetheless, the condition for an assertion such as "She is a witch" to be a metaphor and not a lie is that both the speaker and the hearer are not on a witch hunt and do not believe in witches.

[23] For an recent account of metaphor founded on possible world semantics cf. Hintikka/Sandu 1994.

[24] But both outstanding interactionist accounts of metaphor (Kittay 1985, Indurkhya 1992) lack profound examinations of similes and metaphors. Although Kittay's great study provides the most complex linguistic account of word-field-structures for metaphors, which also holds for similes, she does not succeed in providing a convincing transition to epistemological issues; eventually she pleads for the so-called epistemic access, referring to R. Boyd's (1979) famous contribution to the issue. Indurkhya assumes that metaphors and similes are perfectly congruent, which is due to the fact that in his epistemological approach to metaphors the linguistic means play absolutely no role: his analysis of metaphors proceeds only by analogy to cognition processes. Thus, he unwittingly endorses the general trend in the epistemology of metaphor which is to reduce tropes to mere cognitive functions, assuming tacitly that they are purely conceptual, having no semiotic body.