|
|
Questions and Summary of
"Fake Guns and Stone Lions: Conceptual Blending and Privative Adjectives"
Seana Coulson and Gilles Fauconnier
"Menendez Brothers Virus: Blended Spaces and Internet Humor"
Seana Coulson. 1996. In A Goldberg, ed., CONCEPTUAL STRUCTURE, DISCOURSE and
LANGUAGE. CSLI Publications.
discussed by Elena Dapremont
dapremon@ling.ucsd.edu
"Fake Guns ...
This paper contrasts Bradley Franks' 1995 sense generation treatment of
privative adjectives with a conceptual blending (Coulson 1997; Fauconnier and
Turner 1996) approach to the same construction. Privative adjectives are
defined such that, for a given adjective A and noun N, the claim "No AN is an
N." is necessarily true (Kamp 1975). The goal of the paper is to demonstrate
that conceptual blending is better suited to dealing with concept combination,
of which privatives are a serviceable example.
Coulson and Fauconnier characterize conceptual blending as a theory where
established inputs are combined to yield hybrid frames (blends). The blends
inherit qualities from each input as well as possess unique representational
structures of their own. The inputs and the hybrids each constitute mental
spaces. Mapping permits correspondences to be drawn between objects or
elements in the different spaces. The content of any given blend depends on
the contributions from its input domains. Coulson and Fauconnier contend that
the blending process enables speakers to make connections between elements
whose objective properties may be materially different.
In contrast, Franks' sense generation approach is presented as a model which
relie heavily on these objective properties. The in-context meaning (sense)
consists of sets of features, or attribute value structures (AVS). The three
processes involved in sense generation are UNIFICATION (summation of compatible
AVS's), PRIORITY UNION (unification with overrides) and METONYMIC TYPE COERCION
(MTC). MTC's are of two sorts.
MTC with rebuttal negates central features of the head while combining
diagnostic features with the features of the modifier. MTC with undercutting
allows the inheritance of diagnostic attributes while leaving central
attributes unspecified. Finally, there is IMPLICIT ATTACHMENT which unifies
"explicitly evoked senses" with "contextually evoked concepts". This provides
an account for functional privatives, those adjectives that do not always
behave like privatives (cf. "stone" in "stone lion" and "stone" in "stone
bridge").
Coulson and Fauconnier contend that sense generation's constraints on initial
activation of values in the AVS and its appeal to implicit attachment are
inadequate mechanisms for addressing privatives because, in order to understand
the notion of a "fake gun" or a "stone lion", the speaker must go beyond the
presense or absence of objective features. The speaker must coordinate frames
between the actual and the counterfactual and must exploit existing relations
between the concepts involved. They argue that the attribute value listing of
sense generation does not provide for this. On the other hand, the
cross-domain mapping of conceptual blending does. Coulson and Fauconnier
submit that it is the ability of conceptual blending to accommodate the
requisite flexibility of concept combination that makes it the more appropriate
theory for addressing this type of construction.
POINTS FOR DISCUSSION
I agree that a conceptual blending approach offers a more satisfying treatment
of privatives. There are, however, a few points in this paper that I would
have like to see developed further:
1. I think that the "fake 19th C gun" example might have been dismissed too
easily by saying that acknowledging a [made-in-the-19th C] feature does not
explain why this fake gun can shoot real bullets. Consider the Ritz Cracker
recipe for "mock apple pie". The name is not intended to suggest that the
dessert is not a real pie, but to suggest that it is not a real APPLE pie. In
other words, the "No AN is an N." test does not apply unless "apple pie" is
considered as a unit. This seems to beg a comparison between the boundaries of
a blended space and the extension of an AVS.
2. Coulson and Fauconnier state that feature ascription is a side-effect of
blending. Franks says that the top-down factors of schema-based models only
play a "facilitatory role" in his model. He gives priority to a bottom-up
approach, thereby ensuring that context cannot alter basic feature
combinations. I was unclear as to how conceptual blending accounts for the
fact that, even with the "finger gun" of a cops and robbers game, there seem to
be "gun features" (e.g., being a weapon, having a barrel, theoretical ability
to shoot bullets, etc.) that the blend cannot alter if the game is to proceed.
3. Franks argues that the differences between privatives (proper vs.
functional and undercutting vs. rebutting) are a function of the source and
type of defeat of central attributes. Does conceptual blending theory contend
that these distinctions are artifacts of the blending process? If so, what is
it within the process that allows the differences to be as distinct as Franks
seems to suggest?
"Menendez Brothers Virus: Blended Spaces and Internet Humor"
Seana Coulson. 1996. In A Goldberg, ed., CONCEPTUAL STRUCTURE, DISCOURSE and
LANGUAGE. CSLI Publications.
The goal of this paper is to show that analogical mapping does not necessarily
go from source to target as previously assumed. Coulson uses the computer
virus joke to point out that it is the purpose of the mapping that determines
the direction of the transfer. Specifically, in the case of the Menendez
Brothers Virus joke, the purpose is, presumably, to point out the absurdity of
the brothers' defense. This is a schema projection onto the source domain
which Coulson argues is possible because, in a conceptual blending model, there
is a well-developed blended space (the one that defines the characteristics of
a computer virus) which generates the schema that is transferred. Coulson uses
this example to conclude that the analogical mapping process is not inherently
from source to target. Instead, it is simply part of a fundamental cognitive
process which permits the recognition of similarities across distinct domains.
POINT FOR DISCUSSION
Coulson offers a brief discussion on why the MBV joke successfully paints the
brothers as conspirators while a variation on the joke (MBVD) does not equally
successfully paint the brothers as victims. My reading of the MBVD joke is
that it is more successful at painting the brothers as victims if the phrase
"get revenge" is omitted. It seems that this would suggest that the inference
schema transfer can be disrupted by conflicting mental spaces, in this case,
the less victim-like concept of revenge versus the more victim-like desire to
escape abuse. If this is a reasonable interpretation, what are the
implications with respect to the limitations of mapping "back" to the source?
|
