30.2276, Review: Linguistic Theories; Semantics; Syntax: Petruck (2018)
The LINGUIST List
linguist at listserv.linguistlist.org
Fri May 31 19:33:38 UTC 2019
LINGUIST List: Vol-30-2276. Fri May 31 2019. ISSN: 1069 - 4875.
Subject: 30.2276, Review: Linguistic Theories; Semantics; Syntax: Petruck (2018)
Moderator: Malgorzata E. Cavar (linguist at linguistlist.org)
Student Moderator: Jeremy Coburn
Managing Editor: Becca Morris
Team: Helen Aristar-Dry, Everett Green, Sarah Robinson, Peace Han, Nils Hjortnaes, Yiwen Zhang, Julian Dietrich
Jobs: jobs at linguistlist.org | Conferences: callconf at linguistlist.org | Pubs: pubs at linguistlist.org
Homepage: http://linguistlist.org
Please support the LL editors and operation with a donation at:
https://funddrive.linguistlist.org/donate/
Editor for this issue: Jeremy Coburn <jecoburn at linguistlist.org>
================================================================
Date: Fri, 31 May 2019 15:33:20
From: Lucia Busso [lucia.busso90 at gmail.com]
Subject: MetaNet
Discuss this message:
http://linguistlist.org/pubs/reviews/get-review.cfm?subid=36492317
Book announced at http://linguistlist.org/issues/29/29-4135.html
EDITOR: Miriam R.L. Petruck
TITLE: MetaNet
SERIES TITLE: Benjamins Current Topics 100
PUBLISHER: John Benjamins
YEAR: 2018
REVIEWER: Lucia Busso, Università di Pisa
SUMMARY
The volume “MetaNet” presents a collection of papers which document the work
of the first research project on the formal treatment of metaphors, the
MetaNet project, based at the International Computer Science Institute (ICSI)
and at the University of California, Berkeley. The papers here collected
originally appeared in Constructions and Frames 8:2 (2016).
The MetaNet project is rooted in Conceptual Metaphor Theory (Lakoff & Johnson,
1980) (henceforth: CMT), and was originally funded by the Intelligence
Advanced Research Projects Activity (IARPA), the Intelligence branch of the
U.S. Defense Department. The interest of the U.S. intelligence community was
primarily in public discourse, especially with respect to the way that
linguistic choices reveal the different possible ways to understand certain
timely social issues, such as poverty, governance, and gun control. MetaNet
incorporates insights from Frame Semantics – in particular from the FrameNet
project (Fillmore, 2012) – CMT, and Construction Grammar (Goldberg, 2006;
Sullivan, 2013) with Corpus Linguistics techniques for the analysis of
linguistic expressions of metaphor in very large natural language corpora. The
MetaNet project developed a formalization for Conceptual Metaphor Theory, and
created tools and techniques for the automatic identification and analysis of
the linguistic expression of metaphor. As above mentioned, the MetaNet project
adopts a frame-based approach to the representation of meaning, directly using
actual FrameNet frames where possible. FrameNet and MetaNet, furthermore, have
similar approaches (both heavily theory-grounded and data-driven): both
research projects are structured based on natural language data, and rely on
additional data to refine and expand the analyses, be they with respect to
frames, lexical material, or metaphors. The MetaNet project is an extremely
helpful and useful resource for all researchers interested in metaphor theory,
in that it addresses the internal structure of metaphors, relations between
metaphors, which are represented as frames in the fillmorean sense, and
networks of metaphors. In the scientific landscape, MetaNet’s identification
and analysis system is unique in that it produces an analysis of linguistic
metaphoric expressions based on its repository of metaphors, and it models
frames and metaphors, as well as relations among them.
The contributions of this volume present shared interests and methodology, and
report on the work of the MetaNet project
(https://metaphor.icsi.berkeley.edu). Namely, all papers in the collection
exploit MetaNet repository and its computational pipeline for automatic
metaphor recognition and extraction, aiming at a description of metaphorical
language that integrates Frame Semantics, CMT and Construction Grammar.
The only exception is Sullivan’s paper, which is theoretical in nature and
provides background information about the theoretical and conceptual
underpinnings of current work in metaphor research. The MetaNet project
encompasses the study of four languages (i.e. English, Russian, Mexican
Spanish, and Iranian Farsi); however the papers in the current edited volume
only focus on English.
The collection begins with a Preface by George Lakoff, followed by an
Introduction authored by the editor, Miriam R.L. Petruck, which presents the
MetaNet project and briefly summarizes the relevant background information on
the project itself, and its theoretical foundations.
After the general introduction to the volume, the first contribution is by
Karen Sullivan, and is titled “Integrating constructional semantics and
conceptual metaphor”. The paper is not experimental in nature like the
following chapters, but rather deals with the integration of constructional
semantics with conceptual metaphor. While her contribution does not stem from
direct work on MetaNet, Sullivan’s approach to metaphor research (Sullivan
2009, 2013) was fundamental – in the words of all contributors – for the
development of MetaNet’s extraction/identification and analysis system (see
below). Sullivan’s main thesis in this paper is that although CMT is indeed an
extremely valuable general framework, it also has several limitations that can
easily be addressed by integrating CMT with other cognitive linguistics
frameworks. The author exposes the theory’s shortcomings, namely that CMT is
not clear in explaining which aspects of metaphorical language elicit which
conceptual structures, and should be integrated with constructionist
approaches. To further motivate her thesis, Sullivan points out the
limitations of CMT, as the theory’s exclusive attention to conceptual, and not
linguistic, structure. In fact, CMT alone cannot explain a striking fact: if
one of the arguments in a metaphoric expression is manipulated, metaphorical
meaning is no longer valid. Constructional approaches to syntax, in fact, can
be useful in that they provide an explanation of how particular constructional
slots are associated with different functions in evoking metaphor. In the
author’s opinion, Cognitive Grammar is the best fit and the most effective in
this regard. Furthermore, Frame Semantics helps explain how words or phrases
in the relevant constructional slots evoke source and target domains of
metaphor. Though these theories do not yet integrate seamlessly, their
combination offers nevertheless many explanatory benefits, such as allowing
generalizations across metaphoric and non-metaphoric language, and identifying
the words that play a role in evoking metaphors.The author provides a
compelling example for her theory, taken from the conceptual metaphor THEORIES
ARE BUILDINGS. Given a sentence such as “The lawyer built an argument”, the
metaphoric interpretation stands if we were to change the main verb with other
build-like verbs (e.g. “the lawyer constructed an argument”). However,
replacing the noun argument with a building-related noun (e.g. “the lawyer
built a house”), will force the literal interpretation of the sentence.
Sullivan explains this apparent inconsistency by relying on Construction
Grammar. Specifically, grammatical constructions – i.e. the basic building
blocks of language, pairings of a form and a meaning – contribute to the
interpretation of an expression as metaphorical. In the previously suggested
example of the THEORIES ARE BUILDINGS metaphor, the combination of a
metaphoric verb and a non-metaphoric noun is what determines the metaphoric
interpretation of the sentence. Verbs are more frequently used metaphorically
(i.e. they evoke the metaphor’s source domain) than nouns. Sullivan explains
this asymmetry – which is already known in the literature (e.g. Cameron 2003;
Sullivan 2007) – using Frame Semantics: in FrameNet in fact the frame is
evoked by the verb, while nouns constitute referents that fill roles in the
given frame (Ruppenhofer et al., 2016). Additionally, Cognitive Grammar’s
principles of conceptual autonomy and dependence are also used to support the
author’s claim (Cameron, 2003; Sullivan, 2009) that metaphor analysts should
take into consideration a wider range of notions and tools to better account
for metaphorical expressions.
Elise Stickles, Oana David, Ellen Dodge, and Jisup Hong author the second
contribution “Formalizing contemporary conceptual metaphor theory: A
structured repository for metaphor analysis”. The paper describes the first
endeavour to provide a standard formalization for CMT as instantiated in
MetaNet’s repository of metaphors. In other words, this work begins to address
the “fil rouge” of the volume, that is the goal of bringing rigor and a means
for validation as well as replication to CMT. Specifically, the paper
describes the formalization of CMT and its implementation in a structured
metaphor repository. In fact, metaphor analysis needs the formalization
provided by the MetaNet project in that it develops the internal structure of
frames and relations between frames, as based on an Embodied Construction
Grammar framework (Feldman et al. 2010). In turn, given the already mentioned
dependence of CMT on Frame Semantics, frame structure and frame relations
inform the structure of metaphors and relationships between them.
Additionally, the contribution renders explicit the hierarchical nature of
metaphors and frames: that is, inferential information which begin with (and
as) embodied conceptual primitives is inherited throughout the network. By
providing formal and data-driven representation of the structure of metaphors
and the relations between them, the author also address the long-standing
criticism of metaphor research in the cognitive linguistic tradition,
specifically the reliance on prose description of metaphors and the
insufficient explication (and understanding) of the taxonomic nature of
metaphoric systems.
The following contribution, “Cascades in metaphor and grammar: A case study of
metaphors in the gun debate”, is authored by Oana David, George Lakoff and
Elise Stickles. The paper lays out the principles of the interaction among
metaphors, frames, and constructions in a metaphor network, by thoroughly
describing the case study of U.S. contemporary gun debate discourse. This
discourse, the authors argue, is rife with metaphors that manifest vast
numbers of novel metaphorical expressions. The paper makes use of the
computational pipeline to automatically extract metaphors from texts, and
specifically focus on an important feature of the hierarchical organization of
information that MetaNet uses in its automatic metaphor identification system.
Namely, MetaNet’s ontology of metaphors is organized in terms of metaphor
cascades, that is, “pre-existing packages of hierarchically organized primary
and general metaphors” that co-occur. This type of organizational structure
allows information from one level of the structure to flow (or cascade) to
other levels of the structure. The contribution sets to describe the
architecture of metaphor-to-metaphor relations built into this system, and
successfully provides a thorough examination of the metaphors and the
inferential structures underlying the discourse of gun control in the United
States. In particular, the authors highlight differences and similarities in
conceptual and metaphorical structures in the two sides of the debate:
Government Control versus Individual Control.
Another case study concerning metaphors in political and social discourse is
presented by Ellen K. Dodge, the author of the following contribution, “a deep
semantic corpus-based approach to metaphor analysis: A case study of
metaphoric conceptualizations of poverty”. This paper analyses metaphorical
expressions about poverty, using data taken from the Gigaword corpus (Graff &
Cieri 2003). Analyses of the dataset reveal that specific metaphor phrases
that are used in the discourse on poverty are instances of more general
systems of metaphor. More specifically, metaphorical language involving
poverty tends to draw from two very different broad source domain networks,
each supporting different types of inferences about poverty, its effects, and
possible ways to reduce or end it. Namely, the first frame family which
constitutes one of the two most frequently used source domain is the
Location/Motion network, which includes the frame families of Position (“live
in poverty”), Translational Motion (“fall into poverty”), and Motion
Impediments (“the trap of poverty”). The second one is instead the Physical
Harm network, which contains the frame families of Disease (“alleviate
poverty”), Physical Combat (“poverty attacks children”), Endangerment (“the
threat of poverty”), and Harmful Encounters (“suffer from poverty”). Dodge’s
paper show how the inferential discourse underlying the two different source
domains result also in opposite view of the problem of poverty and its
possible solutions: while the former network (Location/Motion) individuates
poverty as essentially a problem of individuals, the latter (Physical Harm)
conceptualizes poverty as a societal problem, that must be addressed not by
single individuals but by society as a whole. Such work contributes to a deep
understanding of metaphorical language implications in discourse, and also
demonstrates the value of corpus-based and constructional metaphor analysis.
The last paper in this collection describes and illustrates MetaNet’s
computational extraction system for finding instances of metaphoric linguistic
expressions in large corpora of natural language data. The contribution is
authored by Jisup Hong and is titled “Automatic metaphor detection using
constructions and frames”. The metaphor identification system illustrated in
the present contribution applies theoretical principles from construction
grammar, frame semantics, and recent developments in CMT. The author
analytically describes the system’s iterative process to identify metaphoric
uses of language, based on constructional pattern matching. The system
evaluates the likelihood of an expression being metaphoric by using a weighted
confidence score method. New occurrences of metaphorical language are compared
to the MetaNet’s repository, to determine if they are instances of conceptual
metaphors already present in the database. For any example of conceptual
metaphors found by the system, the output is an annotation of source and
target domain constituents. The iterative nature of the process determines
both the system’s refinement as well as improvement to the repository.
Additionally, because of its theoretically principled design, the extraction
system described in the paper provides useful feedback to linguists about the
accuracy of frame and metaphor analysis in the network. The system is shown to
achieve relative success in identifying metaphorical expressions for a range
of target domains from large corpora, and is a promising tool for any
corpus-based study of metaphor.
A brief epilogue by Eve Sweetser wraps up and concludes the volume.
EVALUATION
The present volume constitutes an extremely valuable contribute both to the
study of conceptual metaphor theory and to constructional approaches to
language. It represents the first research project that cohesively
incorporates Frame Semantics, Conceptual Metaphor Theory, and Construction
Grammar with Corpus Linguistics techniques. The MetaNet project is the proof
that research on language is a fundamental tool of investigation of our
reality, be it cultural, political, psychological or societal. This book is an
essential read for any researcher interested in metaphorical language and its
significance in our Weltanschauung, not only from a linguistic perspective,
but also from literary, poetic, cognitive, or computational points of view.
The volume brings together six papers from different authors, all members of
the MetaNet project, who present valid and noteworthy contributions on CMT and
its integration with constructional approaches of language and
state-of-the-art computational and corpus linguistics techniques. The
structure is well-organized and coherent. The introduction is extremely clear
and informative, and summarizes in an efficient way the purposes and the
content of the book, and gives useful information on the MetaNet project and
its aims. The paper by Sullivan links CMT to Usage based theories of language,
contributing to delimiting a more formal and quantitative approach to metaphor
analysis. At the same time, the paper provides additional cues to the
existence of constructions as cognitive realities in speakers’ mental lexicon,
corroborating research in construction grammar and cognitive linguistics.
Sullivan’s contribution is excellent in all respects, and provides the reader
a clear and thorough theoretical basis which lays the groundwork for all
subsequent chapters.
The following paper by Stickles et al. represents an in-depth and fine-grained
review of MetaNet frame-to-frame and metaphor-to-metaphor relations. The
authors make extensive use of multiple real examples, which facilitate the
comprehension of both theoretical foundations and computational applications
of MetaNet and CMT. The inferential structure captured by the MetaNet
repository is accurately explained and addressed.
Following these two “introductory” chapters, two contributions on social
discourse fully demonstrate the type of analysis that the MetaNet repository
and computational pipeline allow. The first of these two more data-driven
contribution is the paper by David et al, which describes with great detail
and accuracy a fine-grained computational analysis of metaphoric cascades (see
above), using a compelling case study on metaphorical language in the debate
about gun violence and gun possession in the United States. The debate of gun
control is the perfect test ground to analyse the importance and
ubiquitousness of metaphorical language in everyday discourse, and its
capability to express (and influence) thought. The second paper which
illustrates the fruitful application of MetaNet is the one by Dodge, which
describes a case study on poverty metaphors. The paper successfully shows that
the representation of individual metaphors and frames as parts of larger
conceptual networks facilitates analyses that aim to grab local details and
larger patterns of metaphor use. However, the description of the corpus used
for the analysis is – in my opinion – lacking, differently from the thorough
description of the data in the previous chapter. Namely, no information is
given on the types of texts that compose Gigaword –the corpus employed for the
paper. Given the fact that the analysis suggests that the two different
metaphors used in poverty discourse each support an opposite inference about
causes and effects of poverty, it is crucial to better understand which types
of data were used. Notwithstanding this issues, the overall quality of the
contribution is undoubtable.
The concluding chapter by Hong is an extremely technical methodological paper,
which sets out to illustrate the innovative computational pipeline that allows
for the bottom-up metaphor detection system. The pipeline incorporates
Semantic Web languages to automatically retrieve metaphors in texts. The paper
outlines a computational system which represents a useful test-bed – both
theoretical and analytical – to reveal the deductive consequences of frame and
conceptual metaphor analysis. The contribution is very technical in nature,
the most specialized in the volume, and may require some particular attention
from the non expert reader. Notwithstanding the technical nature of the
contribution, the author employs a very clear writing style and provides
detailed explanations and examples, which allows also the most naïve reader to
fully understand the sophisticated system reported in the paper.
In conclusion, “MetaNet” is an outstanding volume that – as George Lakoff
eloquently puts it in the Preface to the volume – “is THE must-read collection
of papers if you have any interest in the deep semantic computational
processing of huge collections of linguistic data”.
REFERENCES
Cameron, L. (2003). Metaphor in Educatinal Discourse. London: Continuum.
Feldman, J., Dodge, E., & Bryant, J. (2010). Embodied Construction Grammar. In
B. Heine, & N. Heiko, The Oxford Handbook of Linguistic Analysis. New York:
Oxford Univerisity Press.
Fillmore, C. (2012). Encounters with language. Computational Linguistics, 38
(4), 1-18.
Goldberg, A. (2006). Constructions at work: the nature of generalizations in
language. Oxford: Oxford University Press.
Graff, D., & Cieri, C. (2003). English Gigaword. Philadelphia: Linguistic Data
Consortium.
Lakoff, G., & Johnson, M. (1980). Metaphors we Live By. Chicago: University of
Chicago Press.
Ruppenhofer, J., Ellsworth, M., Petruck, M., Johnson, C. R., & Baker, C. F.
(2016). FrameNet II: Extended Theory and Practice. Berkeley: ICSI.
Sullivan, K. (2009). Grammatical constructions in metaphoric language. In L.
T. B., & K. Dziwirek, Cognitive Corpus Linguistics (p. 57-80). Frankfurt/Main:
Peter Lang.
Sullivan, K. S. (2013). Frames and constructions in metaphoric language.
Amsterdam: John Benjamins.
ABOUT THE REVIEWER
I hold a PhD from the University of Pisa, Italy, where I currently work as a
researcher at a neurolinguistic project. My research interests span from
psycholinguistics, to pragmatics, to general linguistics. My main approach to
language studies is usage-based. I'm hoping to continue working in academia,
in Italy or elsewhere.
------------------------------------------------------------------------------
*************************** LINGUIST List Support ***************************
The 2019 Fund Drive is under way! Please visit https://funddrive.linguistlist.org
to find out how to donate and check how your university, country or discipline
ranks in the fund drive challenges. Or go directly to the donation site:
https://iufoundation.fundly.com/the-linguist-list-2019
Let's make this a short fund drive!
Please feel free to share the link to our campaign:
https://funddrive.linguistlist.org/donate/
----------------------------------------------------------
LINGUIST List: Vol-30-2276
----------------------------------------------------------
More information about the LINGUIST
mailing list