Presuppositions in Spoken Discourse
Presuppositions in Spoken Discourse
Presuppositions in Spoken Discourse
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Chapter 2<br />
effects of alternative candidates on the w<strong>in</strong>n<strong>in</strong>g output, and a good <strong>in</strong>troduction<br />
can be found <strong>in</strong> Archangeli & Langendoen (1997). Other frameworks generally<br />
treat candidate expressions or <strong>in</strong>terpretations <strong>in</strong> isolation. Blutner argues that an<br />
awareness of alternatives <strong>in</strong> both production and <strong>in</strong>terpretation can <strong>in</strong> part account<br />
for cases where a presuppositional expression is felt to be somehow pragmatically<br />
anomalous.<br />
Basic OT has three key components, Gen, Con, and Eval. Gen generates all<br />
potential candidate outputs for a given <strong>in</strong>put. Con is a ranked set of constra<strong>in</strong>ts<br />
that Eval then uses to evaluate each candidate to determ<strong>in</strong>e the optimal output for<br />
the given <strong>in</strong>put. OT constra<strong>in</strong>ts are soft constra<strong>in</strong>ts, mean<strong>in</strong>g they can be violated.<br />
OT-syntax and OT-semantics are uni-directional forms of OT. OT-syntax takes a<br />
given mean<strong>in</strong>g and then evaluates candidates to determ<strong>in</strong>e the optimal form for<br />
express<strong>in</strong>g this mean<strong>in</strong>g, i.e. tak<strong>in</strong>g the perspective of the speaker <strong>in</strong> production.<br />
OT-semantics takes a form and then evaluates candidate mean<strong>in</strong>gs to determ<strong>in</strong>e<br />
what would be the best <strong>in</strong>terpretation, i.e. tak<strong>in</strong>g the perspective of the hearer <strong>in</strong><br />
comprehension. The evaluation process is represented <strong>in</strong> a tableau where candidate<br />
outputs are listed along with the constra<strong>in</strong>ts relevant to their evaluation.<br />
Blutner (2000) proposes a bidirectional version of OT where form<strong>in</strong>terpretation<br />
pairs are evaluated <strong>in</strong> relation to other candidate form-<strong>in</strong>terpretation<br />
pairs. A superoptimal pair 〈 A, τ 〉 is a pair where a given form A is always optimally<br />
<strong>in</strong>terpreted as lead<strong>in</strong>g to an output τ, and vice versa, e.g. generat<strong>in</strong>g an expression<br />
to reach an output context τ is always optimally done with form A, and vice versa.<br />
Two constra<strong>in</strong>ts cover the ma<strong>in</strong> needs of presupposition <strong>in</strong>terpretation, AVOID<br />
ACCOMMODATION (AVOIDA) and BE STRONG (explanations given below adapted<br />
from Blutner, 2000, p. 209).<br />
44<br />
AVOIDACCOMMODATION (AVOIDA): Count each discourse marker that<br />
must be accommodated as a violation.<br />
BE STRONG : Evaluate pairs 〈 A, τ 〉 with stronger outputs τ higher than<br />
pairs with weaker ones.<br />
The first constra<strong>in</strong>t captures the preference for b<strong>in</strong>d<strong>in</strong>g over accommodation, and<br />
the second captures the idea that speakers will prefer stronger <strong>in</strong>terpretations over<br />
weaker ones, believed by some to be at the root of a preference for higher levels of<br />
accommodation, over lower ones, discussed <strong>in</strong> more detail <strong>in</strong> section 5.4. AVOIDA<br />
is ranked higher than BE STRONG.<br />
Tableau 1 illustrates a simple example of an OT <strong>in</strong>terpretation of a<br />
presupposition triggered <strong>in</strong> the consequent of a conditional <strong>in</strong> an empty context.<br />
This example is based on Blutner’s 2000, example (17), (p. 209). There are three<br />
candidate <strong>in</strong>terpretations, global, <strong>in</strong>termediate and local accommodation for one<br />
form, (here we are not compar<strong>in</strong>g form-<strong>in</strong>terpretation pairs but only unidirectionally<br />
what <strong>in</strong>terpretation would be optimal).