ECONOMY IS A PERSON: A Chinese-English Corpora and Ontological-based Comparison Using the Conceptual Mapping Model

ECONOMY IS A PERSON:
A Chinese-English Corpora and Ontological-based Comparison
Using the Conceptual Mapping Model


Siaw Fong Chung

Kathleen Ahrens

Chu-Ren Huang
National Taiwan University National Taiwan University
Academic Sinica
claricefong6376@hotmail.com kathleenahrens@yahoo.com churen@gate.sinica.edu.tw

Abstract

This paper proposes a corpora-based approach in comparing the Mapping
Principles for economy metaphors in English and Chinese. The Mapping Principles
are validated using an upper ontology (SUMO). This research extends on the work of
Ahrens, Chung and Huang (2003) by examining the ‘economy’ metaphors in Chinese
and English. In Ahrens, Chung and Huang (2003), they proposed to delimit the
Mapping Principle via two steps: First, they used a corpora-based analysis on the
word jingji ‘economy’ to find out the most prototypical mappings in a metaphor
Second, they used an upper ontology (SUMO) to examine whether the mapping
principle is a representation of conceptual knowledge in the ontology. This paper goes
a step further by examining the similarities and differences of source domains in
English and Chinese. Using the Conceptual Mapping Model, this paper looks
particularly into the example of ECONOMY IS A PERSON. This paper observes the
representation of shared knowledge in the source domain in different languages and
explains the similarities and differences by looking into the definition of inference
rules in the upper ontology of SUMO.

Key Words: Corpora, Conceptual Mapping Model, Mapping Principle, SUMO,
ontology

1

---

1.0 Introduction

In the framework of Lakoff and Johnson (1980) and Lakoff (1993), conceptual
metaphors are mappings from a concrete source domain to an abstract target domain.
Lakoff proposes a “general principle” which is “part of the conceptual system
underlying English” (1993:306). Ahrens (2002), however, suggested that this ‘general
principle’ can be formulated in the form of Mapping Principle, an intuitive-based
principle stating the underlying reason for source-domain mappings. These rules were
verified with offline experiments (Ahrens 2002 and Lu 2002) in which they
successfully predicted the reading times for metaphors that follow the mapping
principles and metaphors that do not. Therefore, the ‘general principle’ can be
delimited by providing Mapping Principle, which is specific for a particular metaphor
to reason the mappings between source and target domains.
Ahrens, Chung and Huang (2003) proposed to delimit the Mapping Principle via
two steps: First, they used a corpora-based analysis on the word jingji ‘economy’ to
find out the most prototypical mappings in a metaphor and hence formed the mapping
principle. Second, they used an upper ontology (SUMO
http://ontology.teknowledge.com/) to examine whether the Mapping Principle is a
representation of conceptual knowledge in the ontology. For example, in examining
ECONOMY IS COMPETITION, the knowledge of ‘competition’ has a corresponding

2

---

node with Contest in SUMO and “a War is kind of ViolentContest, which in term is a
kind of Contest” (Ahrens, Chung and Huang 2003). Therefore, the metaphors
ECONOMY IS COMPETITION and ECONOMY IS WAR can be subsumed under
the same knowledge representation. These findings support the mapping principles
that there are specific principles governing the source-target domain mappings.
In this paper, we will focus on one metaphor -- ECONOMY IS A PERSON – and
compare the cross-linguistic data for the source domains of PERSON in English and
Chinese. With these data, we also compare Mapping Principles cross-linguistically in
both English and Mandarin. Two research questions are posed – (a) How similar or
different the metaphor of ECONOMY IS A PERSON represented in English and
Mandarin? (b) Are there differences in the representation of knowledge domains in
English and Mandarin metaphor of ECONOMY IS A PERSON at the upper ontology
level? To answer these questions, this paper adopts a similar methodology adopted by
Ahrens, Chung and Huang (2003) by examining the corpora data as well as extracting
the knowledge representation from SUMO to compare with the corpora data.
However, this paper extends on previous research by examining the mapping in two
languages. By comparing two languages, we can further investigate whether the
similar Mapping Principle is extracting for the similar metaphor in two different
languages. We foreshadow that if a similar metaphor with the same type of

3

---

prototypical linguistic expressions is found in two different languages, the Mapping
Principle should be the same. If the Mapping Principles are the same, the knowledge
representations for both speech communities in describing that metaphor are also the
same. In this paper, we will demonstrate this hypothesis by using corpora analysis of
both Chinese and English metaphor of ECONOMY IS A PERSON.

2.0 Economy and Conceptual Metaphors
Metaphors are present in every day’s language use. Some of these metaphors are
so often used that the speakers are unaware of their metaphoric meanings.
Charteris-Black (2000), for instance, carried out a comparative language analysis of
the Economist magazine and the economist section of the Bank of English corpus.
The results suggested that the metaphoric lexis in the Economist were higher in
frequency than in the general magazines. This suggested that the ESP learners are
dealing with more specific types of metaphors as part of their ‘technical’ register.
Incorporating this idea in teaching, Boers (2000) carried out an experiment
comparing the teaching of economy metaphors to two groups of learners – one with
special attention to the metaphoric meanings and the other with dictionary definitions
of the metaphors. The subjects were the French-speaking university students of
business and economics in Belgium. The targeted items for his experiment were

4

---

overcoming a hurdle, bailing out, weaning off, shifting tack and weeding out. The
different inputs for both groups were claimed to have affected the understandings of
the learners – with the groups shown the metaphoric meanings performing better than
the other group.
However, Boer’s (2000) analysis of the metaphors lacks theoretical criterion in
categorizing the metaphorical linguistic expressions. For instance, the examples of
Health and Fitness (Boers, 2000:139) range from sickly company to an acute shortage.
In addition, the target domain was unstated -- the term storage is ambiguous – i.e., it
could have literally meant the shortage of medicine in some place or shortage of
workforce. In order to define and delimit the target domain, this paper has chosen to
look at economy metaphors appearing with the term ‘economy.’ By doing so, the
target domain can be delimited. In regards of the source domain, we suggest the use
of a single term and avoid overlapping scopes such as ‘Health and Fitness.’
In what follows, this paper suggests the use of the Conceptual Mapping Model
(Ahrens 2002), which provides a clearer theoretical analysis of metaphors.

The Conceptual Mapping Model
The CMM is a model based within the Contemporary Theory of Metaphor (CTM)
(Lakoff and Johnson 1980, Lakoff 1993). It supports the idea that metaphors have

5

---

systematic source to target domain mapping. However, the CMM goes beyond the
CTM by postulating a principle connecting the mapping between the source and target
domains. The CMM can also be used in analyzing metaphors linguistically by
dividing the metaphorical expressions into entities (nouns), qualities (adjectives) and
functions (verbs).
In Ahrens (2002), the metaphor IDEA IS BUILDING was analyzed. There were
five steps to this analysis. These five steps are listed in Table 1:

Table 1: Analysis of IDEA IS BUILDING using the Conceptual Mapping Model
Step1
Given the target domain of IDEA, native speakers generated all items
related to IDEA
Step 2
These generated items were categorized into similar source domains
such as BUILDING and WAR
Step 3
For each source domain, the conceptual real world knowledge was
generated. This was done by asking the following three questions:
1. What entities does the source domain (SD) have?
-- (for BULDINGS: foundation, structure, model, base, etc.)
2. What quality does the SD or the entity in the SD have?
-- (for BUILDING: shaky, high, short, strong, etc.)
3a. What does the SD do?
-- (for BUILDING: to protect, to shield, etc.)
b. What can somebody do to the SD?
-- (for BUILDING: to live in, to build, etc.)
Step 4
Non-conventional expressions generated in Step 1 were filtered out
Step 5
The actual mapping between the target (IDEA) and source
(BUILDING) were compared with what could possibly be mapped in
the real world.


6

---

For the metaphor of IDEA IS BUILDING, Ahrens (2001:279) proposed the
following connection between the source and target domain pairings:

Idea (originally capitalized) is understood as building because buildings
involve a (physical) structure and ideas involve an (abstract) structure.

This connection is called ‘Mapping Principle’ (Ahrens 2001:279), which
specifies the underlying reason for the mapping of source to target domains.

3.0 SUMO Ontology
SUMO (Suggested Upper Merged Ontology) is a shared upper ontology
developed by the IEEE Standard Upper Ontology Working Group. It consists of
concepts, relations and axioms that address a broad range of domains and interests.
All concepts in SUMO are structured in the form of hierarchy with the root of Entity,
which represents the most general concept. The Entity is divided into Physical and
Abstract. These Physical and Abstract entities are then further divided into more
specific nodes.
Applying ontology in linguistics, Niles (2003) suggested that the incorporation
of the SUMO ontology with WordNet allows ontology to be used “automatically by
applications (e.g. Information Retrieval and Natural Language Processing applications)

7

---

that process free text.” The interest of this paper lies in observing the automated
processing of Mapping Principles in the source-target domain mappings in conceptual
metaphors.
In this paper, we demonstrate how SUMO helps delimit the source domain
knowledge of metaphorical mappings. We also want to demonstrate how the source
domain knowledge differs (or show similarities) across languages. In order to
examining the similarities and differences cross-linguistically, the following section
first displays our corpora analyses for economy metaphors in English and Chinese.
These analyses help extracting the Mapping Principles of economy metaphors in both
these languages. The concepts represented by the Mapping Principles will then be
examined using the SUMO ontology. This incorporation of SUMO into our analysis
allows the source domain knowledge (identified in the corpora analyses) to be defined
at the upper ontology level.
The following section first presents the analyses of English and Chinese
economy metaphors.





8

---

4.0 Corpora Data

Methodology

The Chinese data were extracted from the Academic Sinica Balanced Corpus, a
tagged corpus with over 5 million words of Mandarin usage in Taiwan. The URL
address for this corpus is http://www.sinica.edu.tw/SinicaCorpus/. 2000 search results
of the Chinese term jingji ‘economy’ were analyzed for conceptual metaphors.

The English data were extracted from the corpora of the Linguistic Data
Consortium (LDC), University of Pennsylvania. The URL address for LDC is
http://www.ldc.upenn.edu/ldc/online/index.html. From the lists of corpora, term
‘economy’ was searched within the Wall Street Journal 1994, a corpus with the size of
14.3 MB (about 5 million words). This makes the size of both corpora almost the
same for both English and Chinese. For each search, a maximum of 100 pages were
extracted. Each page contains 100 instances. This paper selected the first 5 pages to
look at, which constitutes approximately 500 instances of ‘economy’ in the corpus.
This paper has chosen to delimit the target domain of economy metaphors by
running a search on the term ‘economy’ or jingji only. Other related terms such as
‘currency’ and ‘market’ are not the concerns of this current paper.
For both Chinese and English corpora, all instances were read through and
metaphorical uses of ‘economy’ or jingji were marked manually. A metaphor was

9

---

identified when the term ‘economy’ was expressed using concrete idea. For example,
in the Chinese corpus, occurrences such as jingji chengzhang 經濟成長 ‘economy
grew’ and jingjizhan 經濟戰 ‘economic battle’ were identified as metaphorical
instances because there are the concrete domains of ‘growth’ and ‘war’ in the
description of the economy1. Similarly, for English, instances such as ‘growing
economy’ and ‘sputtering economy’ are identified as metaphorical due to the mapping
of the concrete ideas of ‘growth’ and ‘engine’ in the metaphors. These metaphors were
then collected and categorized according to different source domains (GROWTH
CYCLE, WAR, COMPETITION, etc.) in Chinese and English respectively.

Results

The English corpus data produce a total of 209 recurring economy metaphors.
Comparatively, in the Chinese data, a total of 311 recurring metaphors were found.
The breakdowns of the data are shown in Table 2.








1 In the next paper, we will demonstrate that linguistic expressions such as ‘growth’ and ‘war’ are
definable as metaphors if they are hypernyms for at least one concrete and one abstract concept in the
Wordnet. This incorporation of Wordnet strengthens the automation of the Conceptual Mapping Model
in metaphors processing.

10

---