How Does One Recognize a Metaphor?
So Glucksberg's theory is that metaphor vehicles are read as exemplaric members of a larger, unnamed category. Thus, in parliament fumbles, we read the word fumble as an exemplar representing a more general background concept which shares all its salient properties with fumbling.
The question then is how we know when to take a word seriously, and when we should read it as over-specific. Glucksberg seems to think that certain words (like fumble, gold, time bomb, and fly) have an inherent tendency to lean towards a metonymic reading. He calls this their "dual-reference" property:
There is something unsatisfying about this suggestion: If having a dual reference means having a metaphorical reading, then it is tautological to say that the dual-reference is a cure to read something metaphorically.
One cue to metaphoricity might be the dual-reference function of the metaphor vehicle. In assertions such as no man is an island, people implicitly recognize that the word island is used to refer to an abstract category that includes the concrete referent island as one if [sic] its exemplars. People can also implicitly recognize that the metaphor can be paraphrased into simile form as no man is like an island (as awkward as this might seem). […] I suggest that recognition of the dual-reference function of the metaphor vehicle and the recognition that the assertion can be paraphrased as a simile are cues that an utterance might be intended metaphorically. (p. 46)
The only way this assertion can be given content is if we hypothesize that some words -- for whatever reason -- are always recognized as ambiguous between a specific and a generic reading, while others are not. This leaves us instead with the puzzle of explaining this difference.
The idea that paraphrasing the sentence into a simile should be helpful is even more problematic: It essentially boils down to a Gricean process of metaphoricity-testing, which Glucksberg has just spend chapter 2 dismissing as a viable option. So this cannot be a psychologically realistic description of the process.
Idiomatics
In chapter 5, Glucksberg rightly recognizes that metaphors have varying degrees of compositionality. Unfortunately, he spends a little too much time on playful backward-engineering puns like by and not-so-large according to my sensibilities, and that gives him a tendency to overstate the compositionality of idioms.
However, it is generally true and important that co-occurring words have two competing sets of automated readings, a compositional and a phrasal. It is probably the case that the more we hear the words co-occurring, the more we tend to read them phrasally.
Thus, our associations with the words butterfly are quite different from our associations with margarine insect. Note that this is also marked morphologically (through the lack of space in butterfly).
The problem is that the constituents of a phrase can be very frequent and entrenched words as well. It is very hard to get rid of the concrete pictures associated with kick the bucket even if we've heard the expression hundreds or thousands of times. A bucket is simply so sensually rich a word that there's no blocking it.
This Stroop-like effect may also account for much of the data we have on priming and spreading activation in the brain. We know that even phrases like the bark of a tree will prime you for the word dog. Associations and meanings to thus not always agree.
We, the intelligent hearers of a sentence, need to decide how to handle it, and that may require second thoughts or competition between various readings in some cases.
