Disclaimer / Avis de non-responsabilité
Home > The Graph Poem > Metaphor Detection Paper to Be Presented at ACL 2017

Metaphor Detection Paper to Be Presented at ACL 2017

July 27th, 2017 margento

“Metaphor Detection in a Poetry Corpus”

by Vaibhav Kesarwani, Diana Inkpen, Stan Szpakowicz and Chris Tanasescu (MARGENTO)

to be presented soon at the Association for Computational Linguistics Conference 2017, https://sighum.wordpress.com/events/latech-clfl-2017/accepted-papers/   http://acl2017.org/


“Metaphor Detection in a Poetry Corpus”

by Vaibhav Kesarwani, Diana Inkpen, Stan Szpakowicz and Chris Tanasescu (MARGENTO)

Metaphor is indispensable in poetry. It
showcases the poet’s creativity, and contributes
to the overall emotional pertinence
of the poem while honing its specific
rhetorical impact. Previous work on
metaphor detection relies on either rulebased
or statistical models, none of them
applied to poetry. Our method focuses
on metaphor detection in a poetry corpus.
It combines rule-based and statistical
models (word embeddings) to develop
a new classification system. Our system
has achieved a precision of 0.759 and a
recall of 0.804 in identifying one type of
metaphor in poetry.

1 Introduction
Metaphor is crucial in the understanding of any literary
text. A metaphor deviates from the normal
linguistic usage. It intends to create a strong statement
that no literal text can accomplish. Metaphor
differs from idioms, because one can understand a
metaphor even with no prior knowledge. Here are
examples of metaphor in poetry:
 The hackles on my neck are fear (Wright,
 My eyes are caves, chunks of etched rock
(Lorde, 2000)
Literary metaphor operates not only in the local
context where it appears. It also functions in
the broader context of the whole work or even an
author’s oeuvre, and in the context of the cultural
paradigms associated with a specific metaphor
field (Ritchie, 2013). Contrary to the standard
view, literary metaphor sometimes also maps not
only in one direction (from “vehicle” to “tenor”)
but in two. It thus helps reshape both concepts involved
(Ritchie, 2013, p. 189). In other cases, a
metaphor interconnects two concepts and so only
develops each of them into independent sources of
introspective and emotional stimulation (Ritchie,
2013, p. 193).
Literary metaphor is generally thought to be
more stylistically colourful. It is placed somewhere
at one extremity of a spectrum that
has common-speech metaphor at the other end
(Ritchie, 2013). In poetry sometimes the opposite
is also true. The most unadorned and literal
language can be strongly metaphorical by means
of the symbolic import of whole passages or even
entire poems: a poem or a longer passage figuratively
alludes to an implicit concept. Such is the
case, for instance, of Robert Frost’s “The Road
Not Taken” (Frost, 1962). The poem speaks in
its entirety of a consequential choice made in life,
without apparently deploying any actual metaphor.
Needless to say, it is a type of metaphor possibly
even more difficult to process automatically.


Our current work belongs in the same category
as the “GraphPoem” project (MARGENTO, 2012;
Lou et al., 2015; Tanasescu et al., 2016). The milieu
is the computational analysis of poetry, and
the goal is the development of tools that can contribute
to the academic study of poetry.

MORE at ACL 2017

The paper will be presented as part of the SIGHUM (the ACL Special Interest Group on Language Technologies for the Socio-Economic Sciences and Humanities) section.  SIGHUM groups together researchers interested in computational linguistics applications in all aspects of digital humanities and provides a forum for communication between ACL researchers and other digital humanities communities and organisations.  More here: https://sighum.wordpress.com/events/latech-clfl-2017/accepted-papers/


UPDATE: The full paper has been meanwhile published by the Association for Computational Linguistics and is available online here, http://www.aclweb.org/anthology/W17-2201

Categories: The Graph Poem Tags:
Comments are closed.