This preview shows page 1. Sign up to view the full content.
Unformatted text preview: , observes:
As in so many of the examples reviewed in the book, a striking feature of
the blended construction is its underspecification. Although there are
strong constrains on blending, which I shall recapitulate below, there is no
recipe for knowing what will be projected from the inputs and what will
be projected back. In that respect the system is very flexible. In view of the critique of the feature theory on very similar grounds, it is
difficult to see this flexibility as a theoretical gain of the new approach. It
seems that the only way out of this conundrum is to change the perspective on the issue. Such an attempt, among others, is made in Głaz (2002).
In his description of the meaning of the domain of EARTH, he employs
the dynamic usage-based context-dependent model deriving from Langacker’s (1987, 1991, 2000) approach and supplemented with Fuchs’s
(1994, 1999) dynamic semantics. Głaz comes to the conclusion that
In texts, then, we are dealing with activations of semantic regions. Within
the regions, it is possible to recognize areas of greater salience, easier to
identify and name than others, which can be represented in network nodes.
Nodes also serve as convenient landmarks for identifying textual meanings of the relevant item, although in the majority of cases such meanings
do not correspond to the nodes in a one-to-one fashion
(Głaz 2002: 101). Later he adds that the senses of lexemes should be viewed not so much as
network nodes but rather as open regions in semantic space (Głaz 2002: Chapter I 46 107).16 Semanticists should therefore content themselves with approximate definitions (Głaz 2002: 107). In this way what was regarded as a
weakness of lexical semantic studies can be changed into their strength.
Approximate definitions are not incomplete, because of defective or imperfect lexical analyses, but are a result of the dynamic nature of lexical
Coming back to BT, as the blending process can become rather
complex, Fauconnier (1997: 160) notes that
To understand the sentence in context is to have some idea of the kind of
blend intended. But it may take a lot of elaboration for the speaker and the
hearer to converge on sufficiently similar constructions. And, then again,
there is no need for convergence. The folk theoretical illusion that each
expression of language has a meaning that we all retrieve in basically the
same way allows interlocutors to interact under the impression of mutual
comprehension, when in fact they may be engaged in quite different mental space construction. Although miscommunication and misunderstanding do happen and such
partial convergence may be responsible for it, but to claim that “there is
no need for convergence” seems too radical. Without at least partial convergence of communicated conceptualizations, any social activity would
be doomed to failure, which is not the case.
The proposal presented in Fauconnier (1997) is further elaborated
in Fauconnier – Turner (2002). One of the issues they expand on is the
View Full Document