In a 1992 APA paper, we argued that linguistic and pictorial metaphor rely, at least in part, on different mechanisms. For example, language can explicitly predicate a property to some subject-as when I say "Mondays are murderous"-whereas pictures cannot because the syntactic subject-predicate relation does not explicitly exist anywhere in a picture. Instead, pictures sometimes fuse images of things in order to combine their various properties. Although fusion does not lead to as determinate interpretations as predication, it has the advantage of allowing many objects to be combined simultaneously in a manner that is more felicitous than in language.

Consider this political cartoon. It depicts the one-time mayor of Montréal, Jean Drapeau, in the combined form of an elephant and Montréal's Olympic Stadium (which Drapeau built for the 1976 Olympics at huge public expense, much to the consternation of many of his constituents). It conveys several metaphorical claims simultaneously. One is that the Olympic Stadium was a "white elephant," a turn of phrase that was popular at the time. Another is that Drapeau was, himself, the Olympics, in the much the same grandiose, self-important sense that Louis XIV claimed, "L'etat, c'est moi!" Contained in the picture is also a metaphorical reference to Drapeau's own obesity, by way of depicting him as an elephant. All three metaphorical claims are implied simultaneously in the picture.

The question we asked in 1992 was whether such a comic could be "translated" into an apt verbal form, and if not, why not?. A verbal rendering of the comic can be constructed by simply conjoining descriptions of the various component images, but this somehow fails to capture something of the pictorial version: the Olympic stadium is a white elephant and Drapeau is the Olympics and Drapeau is an elephant and so on. Some form of integration stronger than a simple conjunction of the images is at work in the pictorial version. It is not simply Drapeau and the Stadium and the elephant. We argued that this extra something is a fusion of several distinct images, so that otherwise disparate metaphors can be concentrated, as it were, into a single powerful image.

We did not claim, however, that pictorial metaphors are better tout court than verbal ones. We argued that certain verbal metaphors, which depend on syntactic relations such a predication, or certain logical relations such as negation, are difficult to aptly render in pictorial form as well. For instance, the metaphorical claim "She looks like a million bucks" cannot aptly be put in pictorial form by, say, depicting a wad of bills that has been suitably trimmed or folded into female form. This, we argued, is because the verbal metaphor predicates the (metaphorical) wealth to her looks, whereas the pictorial version only fuses the two.

So, to summarize, according to our 1992 position, pictorial metaphor is distinct from linguistic metaphor in that it can make use of the mechanism of fusion. We argued that there is no parallel to this in linguistic metaphor. We now wish to revise this claim by distinguishing on the linguistic side between verbal metaphors that are generated by syntactic features such as predication and negation, and those that, though verbal, are generated by non-syntactic factors. Among such non-syntactic factors, we now argue, are some that bear a remarkable resemblance to fusion.

For instance, in the conventional verbal metaphor, "Surgeons are butchers," some properties of butchers are predicated of surgeons. In a non-sentential verbal expression, however, such as "rose, wife, death, sadness," we see, instead of predication, what seems to be a fusion of the semantic contents of the individual words into a nevertheless recognizable, unified quasi-metaphorical experience. Having discovered this, we went looking for other apparent instances of fusion at work in language. Since, according to our earlier hypothesis, verbal metaphor typically runs off syntactic relations, the most likely place to find the effects of fusion in verbal metaphor would be verbal expressions in which syntax does not play a large role.

This is the case in what linguist Derek Bickerton (1990) calls protolanguage: a form of verbal expression he believes to have been used by hominids before a fully developed syntax organized verbal expression into full-blown language. In protolanguage, concrete nouns and verbs are strung together in a loose order, perhaps determined in part by pragmatic considerations (such as things that happened later in time being mentioned later in the expression), in combination with gestures to convey simple messages. (e.g., "Me, food, give, food, me, hungry.") Although such expressions can be very evocative, it is well-nigh impossible to distinguish precisely between different messages that contain the same basic vocabulary (like "Og ate the wolf" and "The wolf ate Og"). Not only is protolanguage said to be typical of pre-linguistic hominids, but it can also be found in people who speak a pidgin, or by those who find themselves in situations where they do not speak the local language, but are able to pick up a few of the more basic concrete terms.

This level of verbal behavior is also said by Bickerton to be that achieved by those few unfortunate children who are raised to adolescence in a linguistically-impoverished environment. The best-known example of these is "Genie," the girl found in Los Angeles in the 1970s who had been raised to the age of 13 essentially strapped into a chair in an attic by a psychotic father who never spoke to her. Bickerton also argues that protolanguage is all that has ever been achieved by the chimps and gorillas of the so-called "ape-language" studies.

Typically, humans' use protolanguage with the intent of conveying sentential information. Because the intended meaning of a protolinguistic utterance is often obscured by the indeterminacies created by the lack of syntactic marking, the meanings of the words sometimes effectively fuse together leaving univocal interpretation effectively impossible, but a strong sense of the connotation or "emotional content". In effect, the meanings of the individual terms layer over top of one another instead of forming a coherent structure. The effect of this "layering-over-top-of-one-another" is strikingly similar to that we found with fusion in certain pictorial metaphors. Whereas it hampers the ability to effectively convey conventional linguistic information, what we want to argue in this paper is that the tendency of the terms of non-syntactic forms of expression to fuse can be exploited for metaphorical ends in verbal as well as in pictorial expression.

In fact, one can see this phenomenon of verbal fusion at work in some forms of poetic expression. For instance, consider the following haiku by Basho:

In plum-flower scent
Pop! The sun appears-
The mountain path

Notice how terms, rather than being specifically predicated of each other, seem to fuse together into a single unified "image." This fusion, we suggest, may be the source of the imagistic quality for which haiku is famous. As it turns out, conventional translations of haiku into English actually mask the phenomenon to some degree. Although the translation just given is the conventional one, a more literal translation of the same poem would read:

Plum of scent in
pop sun of appearance
mountain path!
(Aitken, 1996, p. 30)

Occasionally, protolinguistic utterances have an eerily similar quality to haiku. Consider the following utterances by "Genie," the girl raised until the age of 13 without language: "Very sad, climb mountain, or "Father take piece wood. Hit. Cry," (cited in Bickerton, 1990, p. 116).

One of the serious drawbacks of protolanguage, as it is typically used, is that the longer the utterance, the more indeterminacies and ambiguities in interpretation creep in. This limits most protolinguistic utterances to about four words. If one's intent, however, were to exploit these ambiguities for poetic purposes, one might intentionally construct longer utterances to multiply the metaphorically-evocative fusions among its elements. Consider, for instance, the following poem by e. e. cummings (1926):

"next to of course god america i
love you land of the pilgrims' and so forth oh
say can you see by the dawn's early my
country 'tis of centuries come and go
and are no more what of it we should worry
in every language even deafanddumb
thy sons acclaim your glorious name by gory
by jingo by gee by gosh by gum
why talk of beauty what could be more beaut-
iful than these heroic happy dead
who rush like lions to the roaring slaughter
they did not stop to think they died instead
then shall the voice of liberty be mute?"
He spoke. And drank rapidly a glass of water

The terms seem to wash over one another into a patriotic "collage" of words. Contrary to our earlier claim that fusion does not occur in verbal expression, this would seem to be verbal fusion to a tee.

In our 1992 paper, we argued that the impossibility of "translating" pictorial metaphors that rely on fusion into sentential metaphors, and vice versa, is crucial evidence of their dependence on distinct mechanisms. If our present revised claim that some verbal, but non-syntactic, expressions rely on fusion in a way similar to that present in pictorial metaphors, then one might argue that we should be able to "translate" from one fused medium into the other. Consider, however, that by their very nature the meanings of fused metaphors are not entirely determinate, so establishing that "translations" from one medium into the other are "exact" is well-nigh impossible.

A weaker form of confirmation of our new thesis, however, may be derived from the identification of pictorial and verbal representations that seem to fuse together similar images to produce similar effects. One could imagine a recruitment poster for the U.S. Marines consisting of a collage of pictures of a waving U.S. flag, a platoon of soldiers fighting their was up a hill, the Statue of Liberty, a Navy ship firing its guns, a mdeical helicopter evacuating the wounded, all topped by a soaring bald eagle. The overall effect of such a poster would not be unlike that of the e. e. cummings poem above. To give it the same ironic tang as the poem, the whole collage might be placed in a thought cloud of a slightly drunk, slightly paunchy ex-military man, sitting at attention and saluting on a bar stool. It would seem that approximately the same connotation can be achieved by both verbal and pictorial fusion.

To take another, somewhat contrasting, instance, compare the overall effect of the e. e. cummings poem to Picasso's Guernica. Although they certainly cannot be said to mean the same thing at all, our impression is that they fuse images of war in ways that are not dissimilar. No element of each is, strictly speaking, predicated, of any other. They are integrated into a complex set of relations that result in an unified experiential Gestalt, of sorts.

All this gives us a potential handle on the perennial problem of how it is that pictures are able to be communicative in a relatively univocal way without the benefit of a specific syntax. In protolanguage, utterances are understood despite their formal ambiguity because the interpretation is constrained by background knowledge and pragmatic inference patterns common to both the speaker and listener. For instance, "Genie" is reported to have said "Applesauce buy store." This could be interpreted in any number of ways (e.g., The applesauce buys the store?, Buy the store from the applesauce?), but because we know about applesauce, stores, and how one might acquire one from the other, the intended message is clear enough that almost anyone would get it.

Something similar seems to be going on in pictures. From our perspective, a painting such as Guernica does not have a specific syntax which tells us how it is to be interpreted. Because we share, however, a great deal of background knowledge (e.g., such as what war is generally, and what bombing is) and pragmatic inferential abilities (e.g., about the likely result of a conflict between modern war technology and traditional society), we are able to achieve a consensus with respect to the message of the painting despite a great deal of fusion-and thereby a great deal of formal ambiguity-among the elements within it. On the other hand, because of the fusion involved, the painting can involve many more combined images, resulting in a much richer experience than would otherwise be possible. Such messages, of course, need not be literal; they can be metaphorical as well. Fusion, in combination with the constraints of knowledge and pragmatics lead to a common understanding of the metaphor that is presented, just as they can with a "literal" depiction.

This leads us to a general speculation: it is quite possible that the cognitive mechanisms that allow for the successful use of protolanguage in appropriately constrained situations, are the same as those that allow for the successful expression of specific semantic content in pictures, regardless of whether that content is literal or metaphorical. The ambiguity that is dangerous in protolanguage (as it is typically used) can be exploited in an almost Gricean sense, in both poetic verbal expression and metaphorical pictorial expression.

To summarize, we have argued that our earlier position concerning the difference between verbal and pictorial metaphor was too limited. Fusion can occur in verbal expression, though it is most apparent only when the usual syntactic apparatus of language is attenuated or stripped away altogether. This is precisely the case in protolanguage. Although protolinguistic utterances are usually kept short in order to avoid just this effect, it can be accentuated to poetic or other figurative effect by lengthening the utterances, as in haiku and some modern Western poetry. Although the evidence is, and perhaps must remain, impressionistic, it seems that the kind of fusion that takes place in these verbal contexts is similar to that which takes place in pictures employing fused images.

This is a symposium on pictorial realism and rhetoric, two topics about which I have said nothing very specific thus far. Going into the US presidential election season, however, I think the implications should become clear enough. As you watch the conventions, the speeches, the video tributes, and the advertisements, think about what is being truly asserted and what is merely being implied by the fusion of images-both verbal and pictorial-of patriotism, trustworthiness, and family. Although the Enlightenment promised us a Government of Reason, it has, perhaps, given us Elections of Poetry.

Aitken, R. (1996). A Zen wave: Basho's haiku and Zen. New York: Weatherhill.

Bickerton, D. (1990). Species and language. Chicago: University of Chicago Press.

Green, C. D. & Vervaeke, J. (1992, August). Pictorial metaphor. Paper presentation at the annual conference of the American Psychological Association. Washington, D.C.