Lexical and supra-lexical underspecification rooted in a dm-based by theoryman

VIEWS: 22 PAGES: 2

									    Lexical and supra-lexical underspecification
  rooted in a dm-based theory of word formation

                                 August 3, 2007


    We present a formalism for underspecification of lexical meaning. The formal-
ism permits kinds of underspecification for which to our knowledge there exist
no competing proposals in the current literature. The underspecified meaning
representations are based on an account of lexical structure and meaning which
follows the lead of d(istributive) m(orphology) in constructing words from ‘roots’,
according to certain general principles ([2], [1]). A syntax-semantics interface
derives logically transparent meaning representations from the root-based word
structures. In many cases of word forms with distinct meanings, it is possible to
define a common root-based semantic representation for these meanings, which
is underspecified with regard to the semantic differences between them.
    When semantic representations are constructed for sentences that contain
such words, their underspecified word meanings will become parts of the sentence
representations. These underspecified sentence representations can then be used
as input to underspecification resolution algorithms that make use of various
sorts of contextual information. Or they can be left as is, in those cases where
resolution is not needed,
    The formalism has been implemented as an extension of standard UDRT
([3]). In this way the mentioned lexical underspecifications will co-occur with
the kinds of underspecification that are already represented within UDRT. This
makes it possible to account for the interaction between these different types of
underspecification during meaning construction and resolution.
    In this contribution we focus on German verbs. An example of the lexical
underspecification we have in mind is the verb trocknen (dry). trocknen (and
likewise dry) occurs in an intransitive, inchoative as well as a transitive, causative
version. Our account proposes a lexical representation of the common core that is
shared between these two versions. Further specification of a particular meaning
may be triggered by the clausal context. If only a nominative NP is present
we get the inchoative reading and in the presence of both a nominative and an
accusative NP we get the causative one. The formation of ung-nominalisation and
past participle operates mainly on the common core and preserves the ambiguity,


                                          1
i.e. Die Trocknung des Lackes ist erfolgt and Die Socken sind getrocknet. Only
if arguments are given explicitly we get either the agentive or the non-agentive
reading (Die Trocknung des Lackes durch Infrarotstrahlung ist erfolgt. and Die
                          o
Socken sind mit einem F¨n getrocknet.).
    The representation of ung-nominalisations is in addition underspecified wrt. a
range of sortal readings and at the same time specifies the possible disambiguation
contexts. Die Lieferung, for example, may have an eventive reading or a material
object reading. If the internal argument of liefern is realized by a genitive NP
only the eventive reading survives. In case the local context does not induce any
disambiguation the underspecified representation is kept and allows for potential
disambiguation at some later stage of sentence or discourse interpretation.
    The account of lexical structure not only has advantages wrt. the underspeci-
fied representation of meanings. It also allows predictions as to which verbs have
ung-nominalisations and/or allow for target state readings of their past partici-
ples and nominals. These predications can be made on the basis of sublexical
verbal constructions and their associated meanings.


References
[1] Alec Marantz. Objects out of the lexicon: Objects as events. Handout. June
    11,2005.

[2] Alec Marantz. No escape from syntax: Don’t try morphological analysis in
    the privacy of your own lexicon. In Penn Working Papers in Linguistics,
    volume 42, pages 201–205, 1997.

[3] U. Reyle. Dealing with ambiguities by underspecification: Construction, rep-
    resentation and deduction. Journal of Semantics, 10:123–179, 1993.




                                        2

								
To top