[PLing] Jezik & Linguistic Colloquium, October 22 @ 1pm

[Jezik&Linguistics Colloquia]

Dear all,

We are pleased to invite you to the next Jezik & Linguistics Colloquium
event, which will take place on *Thursday, 22 October, at 13:00 CEST.*  *Nina
Haslinger *(ZAS Berlin) will give a talk titled "*Disentangling sources of
oddness in Hurford disjunctions"*. You can join us on Zoom
<https://ungsi.zoom.us/j/65235206979?pwd=qM8o64uz8HrLMUN6wMV0f0bmMczynj.1>.
We hope to see you there!

Best,

Arne Goelz, on behalf of the linguistics community at the University of
Nova Gorica

*Abstract:*


*Disentangling sources of oddness in Hurford disjunctions*
So-called Hurford disjunctions (HDs), in which one disjunct entails the
other (Hurford 1974), are known to differ radically in acceptability. As a
first approximation, it seems that cases like (i) involving taxonomic
alternatives are odd, while examples like (ii) involving scalar
alternatives are fully acceptable (Gazdar 1979).

(i) # Ann went to France or to Paris.
(ii) Ann did some or all of her homework.

Many proposals relate the oddness of (i) in some way to truth-conditional
redundancy, but this raises the question of why (ii) is not perceived as
redundant as well, given the simpler alternative (iii).  The possibility of
strengthening *some* to 'some but not all' even in embedded positions is
commonly thought to play a role (e.g. Chierchia, Fox & Spector 2012). But
Meyer (2013, 2014) notes what I’ll call the "strengthening/redundancy
dilemma": Even an LF with embedded strengthening as in (iv) is globally
redundant, since the full disjunction is equivalent to the *some*-sentence
within the scope of exh. So the availability of embedded strengthening does
not fully explain why a cooperative speaker would prefer (ii) over (iii).

To address the strengthening/redundancy dilemma, Meyer has proposed that
(ii) has an LF that is non-redundant due to an ignorance inference, i.e.
the inference that the speaker does not know whether Ann did all of her
homework or only some. To derive the oddness of (i), she proposes that the
mechanism responsible for ignorance inferences produces a contradiction
when the disjuncts stand in a contextual, but not in a logical entailment
relation.

(iii) Ann did some of her homework.
(iv) [exh [Ann did some of her homework]] [or [Ann did all of her homework]]

The contribution of this talk is twofold. In the first half, I argue that
contra Meyer (2013, 2014), contextual as opposed to logical entailment is
not a crucial factor in the oddness of (i). Instead, I argue, the decisive
factor is whether the weaker disjunct can be strengthened to exclude the
stronger one given independently motivated ways of resolving the symmetry
problem (e.g. Fox & Katzir 2011) within an alternative set that includes
both disjuncts. Further, as already pointed out in Fox (2014), (ii) is
acceptable even in the absence of ignorance inferences.  If correct, this
means that Meyer’s way out of the strengthening/redundancy dilemma does not
make the right cut between the good and the bad cases, and the dilemma
rears its head again.

The second half of the talk provides a different answer to the
strengthening/redundancy dilemma, based on the idea that sentences like
(ii) are in fact globally truth-conditionally redundant, but such
redundancy does not automatically result in oddness if there is a Manner
principle favoring the redundant sentence over its simpler alternative. One
such principle, I argue, is that expressions that are "more precise", in
the sense of having less semantic variability depending on the QUD, are
preferred over their "less precise" alternatives (Haslinger 2024). Suppose
that whether embedded strengthening takes place is not a matter of the
syntactic presence or absence of exh operators, but rather a matter of the
QUD, with the strengthening operation itself being obligatory (e.g. Bar-Lev
2020). Then a sentence like (ii) is more precise than (iii) insofar as,
while (iii) can be strengthened to 'some but not all' or not depending on
the QUD, there is no QUD on which (ii) has a 'some but not all' inference.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.univie.ac.at/pipermail/pling/attachments/20261019/d30372e3/attachment.html>