Fauconnier: "Generalized integration networks" (2009)

I have quickly skimmed Gilles Fauconniers contribution to New Directions in Cognitive Linguistics (2009).

I still think he is right in pointing out that constructs like My surgeon is a butcher cannot be explained in terms of one-way mappings. I also still think that his theory doesn't have as much to offer as he himself seems to think.

Fairytale Science: Against a "Fallacy"
Somewhere in Fauconnier's mind, there must be a tiny shred of fear that his theory might be closer to psychoanalysis that to cognitive science.

This can be seen in that he explicitly takes on the following "fallacy":

A wide-ranging cognitive operation purports to explain "everything"
Corollary: Such an operation explains "too much" and is unconstrained (p. 149)

Of course, calling something a fallacy doesn't make it so. And indeed, he doesn't have any real argument that can alleviate this worry.

The closest we get is an allusion to "twelve years" of "convergent evidence" which is not followed by any reference (p. 149).

Fairytale Science: "Precise" But Not "Rigorous"
He later reiterates that conceptual integration "is a precise meaning construction operation" (p. 158), but he also admits that the analyses have to be done on a case-by-case basis:

precise analysis of the generalized networks needs to be done for any observed data, just as chemical analysis needs to be done for any unknown chemical, without adding new elements of new principles to chemistry. (p. 158)

Thus, only the skilled analyst can uncover what cognitive operations a reader must perform in order to understand a given piece of surface language.

The consequence is, he adds in a footnote, that

[...] notions like metaphor, metonymy, analogy, and counterfactual, applied to surface products, elude rigorous definition. (p. 158)

On his account, these notions are thus theory terms, embedded in his own theory.

Fairytale Science: "We Have Indeed"
Even more ridiculously, he quotes (p. 150) a single remark Feldman makes in the conclusion of his book (p. 338), calling Fauconnier and Turner's project "a bold attempt to explain much of mental life."

While "bold attempt" does not usually translate into "unqualified success," Fauconnier does not seem to have a problem reinterpreting the remark as a full-fledged endorsement of his theory:

The proper way to understand this comment is that we (Turner and Fauconnier) have indeed shown that conceptual integration plays a necessary role in human mental life as evidenced by the surface products of particular interest to humans. (p. 150)

I'm not kidding -- he actually writes this.

Showcase Semantics: A New Example
In order to demonstrate his theory, Fauconnier considers the a sentence from a news story about a new California law prohibiting smoking in certain bars. The news piece playfully combines the actual cigarette smoke with the cartoon cliché of a "smoking angry" person:

[...] hard-core smokers [...] were so angry that if they had been allowed to light up, the smoke would have been coming out of their ears. (p. 150)

This example contains two of Fauconnier's favorite phenomena: A counterfactual statements and a two-way interaction between the literal and the metaphorical level.

Showcase Semantics: Putting Pieces Together
His analysis is summarized in the figures on pp. 155 and 157. The gist is this:

1. The conventional ANGER image (or "network") is combined with a stereotypical SMOKER situation to combine an ANGRY SMOKER image which include a double (literal/metaphorical) role for the smoke.
2. The NO SMOKING and the SMOKING ALLOWED situations are combined to form a counterfactual smoking situation qualitatively different from the simple SMOKING ALLOWED situation. He calls this counterfactual the SMOKER'S ZOLOFT SPACE.
3. The ANGRY SMOKER and the SMOKER'S ZOLOFT SPACE are combined to produce the hypothetical smoke coming out of the smokers' ears.

After the Smoke Clears: Fauconnier on Fauconnier
He concludes his analysis with, if I may use my own blend, "a Rumsfeld":

Is the newspaper statement contradictory or unintelligible? Clearly, no. [...] Is the piece of data a "blend"? Clearly, yes, and indeed a visible one. (p. 157)

Shortly after that, he also assures us of the psychological reality of his speculative account (by means of a speculative argument):

Is attested data of this kind cognitively significant? Of course. (p. 158)

His argument seems to be that  because we can understand the sentence, and because he can tell a semantic story about this sentence, his story must be true (p. 158).

McGlone: "Concepts as Metaphors" (2001)

The last chapter in Sam Glucksberg's book Understanding Figurative Language is a super-important essay by Matthew McGlone that summarizes much of the psychological evidence against cognitive metaphors theory in a mere 17 pages.

Circularity
McGlone stresses that linguistic evidence alone cannot decide the matter:

How do we know that people think of theories in terms of buildings? Because people often talk about theories using building-related expressions. Why do people often talk about theories using building-related expressions? Because people think about theories in terms of buildings. Clearly, the conceptual metaphor view must go beyond circular reasoning of this sort and seek evidence that is independent of the linguistic evidence. (p. 95)

He also notes that the strong view of metaphor is incoherent because it would imply that we could not distinguish theories from actual buildings (pp. 94 and 105).

Introspection and post hoc analysis
He goes on to discuss the fact that our semantic intuitions may be an unreliable source of knowledge (p. 95-97).

He provides three arguments for this conclusions, Keysar and Bly's quite ingenious false-etymology experiment (pp. 95-96); a new analysis of a legal argument by Steven Winter (pp. 96-97); and the phenomenon of false etymologies (p. 97).

Winter claims, referring explicitly to cognitive metaphor theory, that the only natural reading of "under the color of law" is "under the (false) appearance of legality" (p. 96). McGlone reports that his subjects in fact think the opposite, so the speculative account was in fact quite misleading in this case (p. 97).

McGlone illustrates the concept of false etymologies with the example Martha is the spitting image of her mother (p. 97).

While an analyst may be tempted to conjure up some "cognitive" motivation for the connection between spit and resemblance, the phrase is in fact derived through an abbreviation of spirit and image. Any "cognitive" story would, in other words, just be another slice of the big baloney.

Note that this argument goes very well together with examples of dead metaphors. If metaphors really were so psychologically real as claimed by the cognitive theory, they should all be completely transparent.

The Lack of Support for Analogical Processing
On the following pages (pp. 99-104), McGlone offers a number of arguments that speak for and against the cognitive theory of metaphor based on psychological experiments.

The first list of arguments against the mapping-based account of understanding is based on his 1994 doctoral dissertation and his 1996 article based on it. These are:

• Paraphrases do not respect conceptual mappings Subjects paraphrase the lecture was a three-course meal as, e.g., the lecture was a gold mine (p. 99).
• Similarity judgements are based on content, not mapping Subjects see the lecture was a three-course meal as equally similar to the lecture was a steak for the intellect and the lecture was a goldmine (p. 99).
• Priming effects rest on content, not mapping As in snack => meal vs. goldmine => meal (pp. 99-100).
• Content-based memory cues work better than mapping-based For instance, "large quantity" is better than "food" as a cue for the lecture was a three-course meal (p. 100).

A finding that seems to support the mapping hypothesis is the following, reported by Nandini Nayak and Raymond Gibbs:

• Subjects prefer text continuations that preserve coherence in terms of mappings So if subjects are given a text with three or four ANGER IS HEAT phrases, they prefer continuations that also applies this mapping to one that applies ANGERS IS ANIMAL BEHAVIOR (pp. 100-101).

However:

• Continuations that preserve mappings do not facilitate reading At least not when these mappings are highly conventional. This was shown using Nayak and Gibbs' own stories (pp. 101-102).

McGlone notes that one should also be careful in the interpretation of experiments of these kinds, since priming effects based on surface words might be conflated with priming effects based on mappings (p. 102).

Unfamiliar or Ambiguous Sentences in the Context of Analogies
He goes on to cite evidence in favor of the following claims:

• Ambiguous sentences can be disambiguated by a mapping prime So the sentence the meeting was moved two days forward has different natural interpretations after the deadline has passed and we passed the deadline (pp. 102-103).
• Novel metaphors are facilitated by relevant mapping primes So Sirens will wail every time they meet is more intelligible as a metaphor about arguments if it is preceded by the novel phrase verbal grenades than the conventional phrase shoot down his arguments (p. 104).

He notes that there is probably some analogical reasoning implies when you hear the sentence Rush Limbaugh's bloated ego gobbled up his integrity and used the airwaves as a toilet (p. 104). There certainly seems to be more pictorial "meat" to it than the more anemic stock phrases typically analyzed in the field.

He comments:

As Bowdle and Gentner (1997) have suggested, the processes used to understand any particular metaphoric expression depend on its conventionality. When an expression is completely novel, it requires different kinds of inferential work than when it is familiar. Thus, the conceptual metaphor view is insufficient as a general account of figurative language comprehension, in part because it does not recognize important processing differences between conventional and novel expressions. (pp. 104-105)

This distinction should be seen on the background that the air of poetic strangeness inherent in creative metaphors can also accompany slightly strange formulations of literally true facts.

Consider for instance the following novel metaphors:

• The soul a prison for the body (i.e., not vice versa)
• Her comment really stepped on an emotional landmine.
• Rush Limbaugh uses the airwaves as his toilet.

Compare these to the following true and literal statements:

• My body has 20 nails.
• I ate the food with my mouth.
• Your house is standing on a planet.

I presume that, in both cases, the phenomenological effect is produced by the higher requirements implied by a novel way of seeing an object.

Naomi Quinn: "The Cultural Basis of Metaphor" (1991)

Quinn's contributions to Beyond Metaphor (1991) argues against the claim that metaphor "constitute understanding," and instead proposes that metaphors "are ordinarily selected to fit a preexisting and culturally shared model" (p. 60).

Thought (Mostly) Structures Metaphor
Here are some more elaborate versions of the quotes in which she presents her ideas:

I will be arguing that metaphors, far from constituting understanding, are ordinarily selected to fit a preexisting and culturally shared model. And I will conclude that metaphors do not typically give rise to new, previously unrecognized entailments, although they may well help the reasoner to follow out entailments of the preexisting cultural model and thereby arrive at complex inferences. I do not want to suggest that metaphors never reorganize thinking, supply new entailments, and permit new inferences; but my analysis will argue that such cases are exceptional rather than ordinary. (p. 60)

Metaphors are usually cherry-picked on the basis of prior understanding:

I want to argue further, and I think quite contrary to what Johnson and Lakoff seem to be saying, that metaphorical systems or productive metaphors typically do not structure understandings de novo. Rather, perticular metaphors are selected by speakers, just because they provide satisfying  mappings onto already existing cultural understandings---that is, because elements and relations between elements in the source domain make a good match with elements and relations among them in the cultural model. Selection of a particular metaphor for use in ordinary speech seems to depend upon its aptness for the conceptual task at hand---sometimes, as we shall see, a reasoning task. (p. 65)

If source domains can be considered and rejected, they cannot dictate out thoughts about the target:

I would like to suggest that the metaphor appears to structure inferences in the target domain, carrying these inferences over from entailments in the source domain, only if it be supposed that the selection of this metaphor is unconstrained. Once it is recognized that choice of metaphor is itself highly constrained by the structure of cultural understanding, then it can be seen that reasoners ordinarily select from possible metaphors those that provide them with a felicitous physical-world mapping of the parts of the cultural model---the elements and relations between elements---about which they are intent on reasoning. (p. 76-77)

This can usefully be compared to Gentner et al.'s discussion (1997) of how Kepler systematically tried out different analogies, assessed their suitability, and sought out their limitations (pp. 22-24, 27-29).

Metaphor (Sometimes) Structures Thought
Quinn does not deny that metaphors may sometimes restructure thought and action. She illustrates this with Hans Selye's reconceptualization of hormonal responses as a general systemic "stress," an example taken from Mark Johnson's The Body In the Mind (1987: 127-37).

She further notes that "[a]s the Gentners (1983) demonstrate with the example of electricity, metaphor is especially likely to organize experience and guide reasoning in just those domains for which there is no other available model" (p. 77).Thinking in terms of lemmings or in terms of water in other words makes a difference when you don't know anything about electricity as such (cf. also p. 59).

Some Comments on Lakoff and Johnson
Although Quinn is generally sympathetic to Lakoff and Johnson's project, she has some problems with their style of presentation:

Their argument sometimes takes the form of a semmingly unqualified claim that metaphor underlies and constitutes understanding. (p. 59)

She notes that

readers of Lakoff and Johnson's published works are likely to go away, as I did after what I thought was a careful reading, with a sweeping interpretation of their claim or at least with some confusion about how sweeping their their theory of metaphor is meant to be. (p. 59-60)

She is also very clear about the ambiguous status of "culture" in the theory:

[...] Lakoff and Johnson are not unaware that culture plays some role in understanding: [...] But culturally constituted meaning has no place of its own beside embodied meaning in Johnson's analysis and no systematically developed or well-articulated place in that of Lakoff. (p. 65)

This certainly also applies to Kövecses invocation of "culture" as a dust-bin concept that can account for any piece of data that falls outside of the theory.

What Lakoff, Johnson, And Quinn Should Have Said
I think it is important to realize that much of the confusion about the status of metaphors as mental object go away once we look at them where they belong, that is, in the context of conversation.

Language certainly governs thought and action, if "language" includes assertions such as tomato leaves are toxic or this bridge is solid. This is not obviously because of a Robinson Crusoe-style individualistic cognition, but because we expect utterances to be benevolent manipulations of our actions. This should be true of metaphor as well.

When the Gentners prompt their subjects to see electricity as a pack of lemmings or as a flow of water, they are essentially signalling to the subjects that this analogy should somehow be helpful. The interesting work occurs as the subjects try to unpack this statement and merge the analogy with their own sense of reasonable behavior.

This involves uncertainty and probably some rounding off the edges as the subjects acquire more experience. It would be a mistake to suppose that the analogy contained the instructions for its own implication, or that a "big baby" could not come to use it more loosely, freely, and proficiently with experience.

Gendlin: "How Philosophy Cannot Appeal to Experience, and How It Can" (1997)

I've read Gendlin's opening essay in the anthology dedicated to his philosophy. I find it fascinating, but also quite wrong-headed in some ways.

Gendlin's point is that we have something like a bodily intuition about phenomena that is well articulated by some words and not so well articulated by others. His claim is that by attending closely to this intuition, we can become more acute observers of our life-world, or so I read him, at least.

Looking For Words
His favorite example of this phenomenon is the poet searching for a suitable line to continue a poem (p. 17). He also cites the practice of rephrasing your point when someone doesn't understand you as evidence that there is "a . . . ." that we can succeed or fail at making other people appreciate (p. 13).

In other words, the point seems to be a call to "return to the phenomena themselves," as the phenomenologists said. "Any text or theory," he explains, "becomes more valuable when it is taken experientially in this way" (p. 40).


He illustrates this kind of thinking with an example from Wittgenstein, not about poetry-writing, but about letter-writing:

I surrender to a mood and the expression comes. Or a picture occurs to me and I try to describe it. Or an English expressions occurs to me and I try to hit on the corresponding German one. Or I make a gesture, and I ask myself: What words correspond to this gesture? And so on. (PI 335; quoted by Gendlin on p. 37)

His conclusion is, with explicit reference to the "postmodernism" in general (pp. 3, 6, 9, 19, 34-35), and Derrida (pp. 8, 35-36) in particular:

People's lives include a great deal that they cannot say in the existing language, but can become able to say. As philosophers, let us stop telling people that they cannot possibly have anything to say that is not already in the public language. (p. 34)

Trying To Succeed
I am skeptical about Gendlin's flirtation with the concept of authentic language for the reasons that Richard Rorty has explained in his wonderful essay on Heidegger's mysticism. Even though there truly is some experience of looking for and finding the right words, it is not clear how serious we should take this experience.

A different way to conceptualize the situation would be in terms of skill. Just like looking for a word, the attempt to accomplish some bodily action can be associated with immense frustration and satisfaction. This is obviously not because the action was there all along in some non-realized form, and we should look at linguistic skill and success the same way.

This also clears up another confusion Gendlin's story, namely: Why would we even want to explicate our intuitions in the first place? Is it for the sake of "truth"? "Authenticity"? "Correspondence"? He touches on the answer with his example of rephrasing your explanations: This is an attempt to bring your hearer into a different state, not an attempt to create a good fit.

This communicative purpose could in principle be achieved with the most ridiculous or outlandish effect imaginable; a completely "wrong" word, a gesture, a cartoon. The success criterion is here change in the state of the hearer, not a relation between words and bodily knowledge.

Another way to say the same thing is that a seller who puts a price tag on a commodity chooses a certain "word" that may or may not have the right effect. This is not because it does not "fit" the commodity, but because it would create the wrong effect in the hearer, such as the belief that the commodity was of a cheap quality, or that it is too expensive.

Yet another similar example occurs when you speak a foreign language. In that situation, you have a gold standard for ease of communication---communication in your native language---and that association can produce a whole lot of frustration as you try to achieve as fine-grained distinctions and precise effects with your impoverished skills in the foreign language.

Lakoff and Johnson on Narayanan (1999)

Here's how George Lakoff and Mark Johnson sum up Narayanan's toy model of English verbal aspect in their 1999 book Philosophy in the Flesh:

Narayanan devised an ingenious way to test whether his model of general high-level motor control could handle purely abstract inferences, inferences having nothing to do with bodily movement. He constructed a neural model of conceptual metaphor and then found cases in which body-based metaphors were used in an abstract domain, in this case, international economics. [...] Narayanan then showed that models of the motor schemas for physical action can---under metaphoric projection---perform the appropriate abstract inferences about international economics. (p. 42)

They conclude:

Aspectual concepts that characterize the structure of events can be adequately represented in terms of general motor-control schemas, and abstract reasoning can be carried out using neural motor-control simulation. (p. 42)

Mark Johnson: "Philosophy's Debt to Metaphor" (2008)

Just a couple of quotes from Johnson's contribution to the Cambridge Handbook of Metaphor and Thought (2008):

Conceptual metaphor is a structure of human understanding, and the source domains of the metaphors come from our bodily, sensory-motor experience, which becomes the basis for abstract conceptualization and reasoning. From this perspective, truth is a matter of how our body-based understanding of a sentence fits, or fails to fit, our body-based understanding of a situation. (p. 45)

Indeed, once you make the links, no specific training is needed:

All theories are based on metaphors because all our abstract concepts are metaphorically defined. Understanding the constitutive metaphors allows you to grasp the logic and entailments of the theory. (p. 51; emphasis in original)

Further, we should recognize the "the crucial role of metaphor in shaping and constraining inference in ordinary mundane thinking" and "the pervasive workings of conceptual metaphor in shaping our conceptual systems" (both p. 28). We are also told that "conceptual metaphors lie at the heart of our abstract conceptualization and reasoning" (p. 51).

Also, "metaphors are based on experiential correlations and not on similarities" (p. 46).

"Cognitive Semantics: In the Heart of Language" (1998)

This is an interview with George Lakoff from the first-ever issue of Brazilian journal Fórum Lingüistico, published by the institute for linguistics at the Universidade Federal de Santa Catarina. It was conducted by Roberta Pires de Oliveira.

The interview is notable for containing both some quite frank and some quite revealing assertions by Lakoff.

It is unfortunately quite badly transcribed. For instance, Pamela Morgan is identified as "Helena Morgan," (p. 106), and Lakoff's paper in Andrew Ortony's book Metaphor and Thought becomes "my paper on Ortony's book" (p. 108, my emphasis).

Below is a collection of some of the more striking quotes from the interview, sorted according to theme.

The Ricoeur connection:

Mark Johnson had studied with Paul Ricoeur. So he knew the Ricoeur tradition and the continental tradition and had come to the conclusion, through working with Ricoeur that metaphor was central to thought. But I wasn't at all influenced by that tradition. What influenced me was the discovery that ordinary, everyday thought and language, and specially ordinary everyday thought, is structured metaphorically. That was the major discovery. (p. 89)

The partial order:

It [= cognitive metaphor theory] says that we have basic concepts that arise from our direct interaction with the world and they are not metaphorical, and then we have metaphorical projections of those to more abstract concepts. (p. 91)

With respect to "big babies" and scaffolding:

Piaget saw that the understanding of, for example, causation came out of a child's dropping things. I think that's correct. But Piaget also thought that if you advance from one stage, then you left behind the other stage. This seems to be false. He thought that there was a higher stage of abstract thinking, and that seems to be false. The details are very unpiagetian when you think of the rest of Piaget's work. (p. 96-97)

The Universal Body:

Yes, there are universal concepts. There are universal metaphors, universal aspects of language, because we all have very similar bodies and our physical experiences in the world are very similar. Those are where universals come from. (p. 98)

Inherent structure:

There's a structure that is independent of any particular metaphor of love. It may not be a very rich one. It may not be a highly structured concept... When you have a lover, a beloved, an emotional relationship, a positive emotional relationship, and lots and lots of types of complex feelings, but it may not be structured enough to reason with then you have lots of metaphors that allow you to conceptualize love in terms of other kinds of experiences. (p. 103)

You have to take into account target domain overrides. [...] The target domain override is a case where the mapping is carried through with contradicting internal structures of the target domain. (p. 107-08)

There's also an extremely embarrassing moment on pp. 106-07, when Lakoff shrugs of a counterexample to one of his claims by saying that a particular set of metaphors is just "linguistic expressions, not mappings," apparently without noticing his own blatant methodological inconsistency. He further goes on to imply that if a metaphor isn't universal, it doesn't really count.

Somers: "On the validity of the complement--adjunct distinction in valency grammar" (1984)

In this article from Linguistics 22(4), Harry Somers surveys various linguistic tests designed to detect the difference between optional sentence adjuncts and obligatory verbal complements. He further proposes to replace this binary distinction by a six-step scale ranging from intergral complements to extraperipherals.

Tests for Argument Type
The most import tests Somers discusses are the elimination test, the back-formation test, and his own "do so" test. Here is an overview of the tests he describes:

• Elimination: If the sentence becomes ungrammatical when an element is eliminated, it is a complement. If it doesn't, then the test makes no conclusion (pp. 509-10).
• Back-formation: If the element can be moved down in an embedded wh-sentence or into a entirely separate sentence, it is an adjunct (p. 511).
• Substitution 1: If the elimination test does not yield a conclusion for a particular verb, but it does for a near-synonym, then we are allowed to transfer the complement status back to the original verb (p. 512).
• Substitution 2: If a change of verbs renders a certain element strange or ungrammatical, then that element is a complement. If it doesn't, no conclusion is guaranteed (pp. 512-13).
• Semantics: If the preposition in a prepositional element cannot be replaced by a near-synonym, then that element is a complement (p. 514).
• for/to status: If an indirect object is most naturally expressed as a to-phrase, it is relatively loosely connected to the verb (although perhaps still a complement); if it is most naturally expressed as a for-phrase, it is a complement (p. 515).
• Questions: If the noun within an element can be referred to with who, whom, or what, it is a complement. If it cannot, it is an adjunct (p. 516).
• "Do so": Whatever elements are semantically included in the referent of the anaphor ...and he did so, too are complements. The rest are adjuncts (p. 516-20).

A Scale of "Complementness"
Somers suggests (p. 524) that we give up the binary distinction between complement and adjunct and instead introduce a six-point scale that measures how intimately connected to the verb any particular element is.

His categories, with examples, are (pp. 524-26):

• Integral complement: put at risk, take care (fundamentally affects the verb).
• Obligatory complement: he wrote me a letter (necessarily implied by the verb).
• Optional complement: Greame caught Steve a salmon (optional, but quite restricted).
• Middle: I gently pushed the button (optional and only weakly thematically restricted).
• Adjunct: I pulled the lever in order to eject (optional and almost unrestricted).
• Extraperipherals: we're all patiently waiting for you, you know. (top-level additions)