CHAPTER 5 GENERAL DISCUSSION
5.4 Speech and Gesture Production
ACTIVITY can be conceptualized via three kinds of sources, SPEECH CONTENT can map to
twokinds of sources. The source OBJECT is the common concept used to reason about TIME,
MENTAL ACTIVITY, and SPEECH CONTENT in daily conversations. Both SEQUENCE and
DEGREE can correspond to SPACE and OBJECT; we find more cases where the two targets are
conceptualized in terms of SPACE. Findings from the present study then offer more empirical
evidence to support the notion that an abstract concept is commonly conceived through more
than one concrete concept and the idea that metaphors based on different sources may profile
different aspects of a concept.
5.4 Speech and Gesture Production
Different theoretical hypotheses about the production of speech and gesture—the Free
Imagery Hypothesis, the Lexical Semantic Hypothesis, and the Interface Hypothesis—are
proposed in previous studies. According to the Free Imagery Hypothesis, the content of
speech will not affect what is encoded in gesture. All the data examined in the present study,
however, involve the gestures that are affiliated with corresponding lexicons.13 The referent
of a metaphoric gesture is not the concrete imagery presented but the abstract concept that is
also conveyed in the accompanying speech. Language and gesture cooperate to express
metaphorical thinking, and the content of speech closely relates to manual movement. While
13 Two metaphoric gestures in the conversational data were produced during silence where the speakers had difficulty retrieving words. Because the referents of these metaphoric gestures were not clear, they were not included in the present study.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the Free Imagery Hypothesis provides a view about the production process of speech and
gesture, this hypothesis is not suitable for discussing the findings based on the analysis taken
in the present study. The present study then puts emphasis on the Lexical Semantic
Hypothesis and the Interface Hypothesis.
To begin with, the temporal patterning of speech and gesture in conveying metaphors is
discussed to examine the theoretical hypotheses. The Lexical Semantic Hypothesis suggests
gestures are generated from the semantics of the lexical items. If a person has difficulty to
produce a word for a concept in language, the production of gesture may help he/she to
search a lexical item for such a concept. Hence, it is claimed that a gesture (both its
preparation phase and stroke phase included) usually precedes the lexical component it
depicts. In Schegloff’s (1984: 275) words, “[t]he critical property of iconic gestures...is that
they are pre-positioned relative to their lexical affiliates, achieving their affiliation by means
other than co-occurrence with them”. Iconic gestures are thought to be generated from the
result of the computational stage of the selecting of the lexical items (Butterworth & Hadar
1989). The Interface Hypothesis, on the other hand, suggests that gestures are generated from
the interactions between speaking and spatial thinking. The hypothesis incorporates the
Growth Point Theory of speech and gesture production: “[t]he growth point is seen in the
gesture stroke, together with the linguistic segment with which it co-occurs” (McNeill 1992:
221). In McNeill’s (1985; 1992) framework, he proposed that a stroke of a gesture lines up in
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
time with the equivalent linguistic unit in speech. The temporal synchronization shows that
speech and gesture belong to the same psychological structure and share a computational
stage. With regard to the current data, results from the two groups of metaphors are combined
and discussed together, because they do not show a statistical difference in the temporal
patterning of speech and gesture. Among the 247 metaphoric expressions, 84.6% of them
include metaphoric gestures synchronized with their linguistic referent. Only 14.2 % of them
comprise metaphoric gestures produced before their associated speech. A small proportion
(1.2%) of metaphoric gestures is performed after the related speech (see Table 19 in Chapter
4). The current data shows that gestures commonly synchronize with their associated speech
in expressing metaphors, and provides evidence to support the Interface Hypothesis rather
than the Lexical Semantic Hypothesis.
Next, the relevant linguistic unit accompanying the metaphoric gesture is discussed.
According to the Lexical Semantic Hypothesis, gesture can help lexical search. If a person
has difficulty to find a lexical item for a concept, he/she may produce a gesture to represent
the idea. The production of such a gesture then helps the person utter the word for that
concept in language. Thus, gestures are thought to be dominated by the computational stage
in which a lexical item is selected from a semantically organized lexicon (Butterworth &
Hadar 1989). The Lexical Semantic Hypothesis stands for the notion that the relevant
linguistic unit to affect the content of a gesture is a single word. In contrast, the Interface
‧
Hypothesis suggests gestures are involved in the process of arranging the spatio-motoric
imagery into informational units suitable for speech production (Kita & Öyzürek 2003). The
informational unit suitable for speech formulation is what can be encoded in a clause in
language. Based on the Interface Hypothesis, the relevant linguistic unit to affect the content
of a gesture can be a unit larger than a single word. The present study sorts the relating
speech of the metaphoric gestures into words or phrases. A word refers to the realization of a
lexeme (Katamba & Stonham 2006), such as
xīnkŭ ‘hard’, zăoshàng ‘morning’, and the
compound
huàzhuāng ‘put on makeup’. A phrase is a group of words, such as tiào nàbiān
‘jump there’,
zhuāng shuĭ ‘filled with water’, shàng cèsuŏ ‘go to lavatory’. Table 20 shows
the linguistic unit of the corresponding lexical affiliates of the metaphoric gestures examined
in the present study. Results concerning the two groups of metaphors are discussed together.
Within the 247 metaphoric expressions, the majority of the lexical affiliates associated with
the gestures are single words (82.6%). Phrases comprise 17.4 % of the lexical affiliates
accompanying the gestures.
Table 20. Linguistic units of the lexical affiliates of the metaphoric gestures
Linguistic unit Group
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Example 13 in Chapter 4 is an instance where the grammatical unit of the lexical
affiliate is a phrase. The speaker complains that a teacher always changes his speech content
abruptly and verbally expressed the metaphor SHIFT OFTHE SPEECH CONTENT IS MOTION
with the statement
zhèbiān jiăng yù tiào nàbiān tiào nàbiān tiào nàbiān ‘(he) talks about this
and then (the speech content) jumps there, jumps there, and jumps there’. Accompanying the
phrase
tiào nàbiān ‘jump there’, his gesture depicts the imagery of the motion to different
places. The gesture not only depicts the manner verb tiào but also the trajectories to the
different places which are expressed by
nàbiān in language. In this case, the information
encoded in the gesture corresponds to the unit lager than a single word. Contrasting to the
prediction of the Lexical Semantic Hypothesis, the relevant unit to influence the content of a
gesture is not obligatory to be a lexical item (a word). A substantial portion of the lexical
affiliates are phrases. This finding is in opposition to the claim of Lexical Semantic
Hypothesis but supports the Interface Hypothesis.
Last, the informational coordination between language and gesture is discussed. In the
Lexical Semantic Hypothesis, gestures are generated from the semantics of the lexical items
in the corresponding speech (Schegloff 1984; Butterworth & Hadar 1989). Thus, the
hypothesis predicts that gestures do not convey the information which is not encoded in the
accompanying speech. In the Interface Hypothesis, gestures are generated from the imagery
representations which interact on-line with the linguistic representations (Kita & Öyzürek
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2003). The Interface Hypothesis then predicts that gesture may encode the information
conveyed in speech or the information which is not included in speech. The present study
examined two groups of metaphors: metaphors realized in both language and gesture; and
metaphors realized in gesture exclusively. In the gesture-only group, a concept is
metaphorically expressed in gesture but literally conveyed in speech. The entity metaphor
HARDSHIP IS AN OBJECT in Example 2 in Chapter 3 is an instance of the gesture-only
metaphors. In the utterance bàba
māmā hěn xīnkŭ ‘father and mother work hard’, the concept
of hardship is literally represented by the predicate
xīnkŭ ‘hard’. Whereas, the speaker’s left
hand forms a cupped shape to represent hardship as an object with boundary. Language
merely conveys the target-domain concept HARDSHIP; on the other hand, the source-domain
concept OBJECT is conveyed in gesture even though this information is not included in
speech. In such kind of expression, linguistic and gestural modalities encode different
semantic contents which are relevant for realizing the metaphorical thought. Concerning the
current data, the gesture-only metaphors comprise over a half of all the metaphoric
expressions and provide considerable amount of evidence for the Interface Hypothesis, which
suggests language and gesture can convey different information. In addition, 7 expressions
where the metaphor in language differs from the one in gesture (c.f., Example 5 in Chapter 3)
are found in the twenty-six conversations examined in the present study. These cases show
that the content of a gesture is not necessary to be restricted by what is conveyed in a given
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
speech. Because expressions with different types of metaphors in language and gesture are
not widely found in the current data, more instances of such expressions are needed to discuss
the collaboration of language and gesture.
The cross-modal expression of metaphor shows that the stroke of a metaphoric gesture
usually synchronizes with its lexical affiliate. Results also suggest that the lexical affiliate of
a metaphoric gesture could be a linguistic unit larger than a lexical item. Regarding the
semantic coordination between language and gesture, the gesture-only metaphors which
contain different information in the two modalities reveal that gesture can encode what is not
conveyed in a given speech. The evidence from the current data then supports the Interface
Hypothesis more, which suggests that gesture is generated from the interface representation
between spatio-motoric information and linguistic information.
5.5 Summary
In this chapter, the findings from the present study are compared with the ones in the
past research. Regarding the metaphor types, the results from the present study partially agree
with McNeill’s (1992) gestural study on narratives. Results from the present data sustain
McNeill’s claim that entity metaphor and orientation metaphor are instantly accessible. On
the other hand, the current data opposes his assertion that Chinese does not use the imagery
of a bounded object as the common source. The source-domain and target-domain concepts
found in the present study are generally consistent with the common sources and targets
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
proposed by Kövecses (2002) in his survey on metaphors in English. Similar to past studies,
the present study finds the metaphors with one-source-to-many-targets correspondences.
Excluding FORCE, which merely has one token in the current data, all the other
source-domain concepts can be employed to conceptualize multiple targets. It also found that
several targets can be conceptualized through various sources. These targets involve TIME,
MENTAL ACTIVITY, SPEECH CONTENT, SEQUENCE, and DEGREE. The present study
provides more instances to support the view that different metaphors may profile different
aspects of a concept.
The theoretical hypotheses about speech and gesture production are discussed as well.
The temporal patterning between speech and gesture in conveying metaphor, the linguistic
unit of the lexical affiliate accompanying the metaphoric gesture, and the semantic
coordination between language and gesture in expressing metaphors are discussed to examine
the hypotheses. The Lexical Semantic Hypothesis suggests that gestures are generated from
the semantics of a lexical item; on the other hand, the Interface Hypothesis proposes that
gestures are originated from the interaction between spatio-motoric and linguistic
representations. Results show that a gesture stroke usually co-occurs with its lexical affiliate,
which supports the Interface Hypothesis. It is found that not all the lexical affiliates belong to
a single word, which contradicts the claim of the Lexical Semantic Hypothesis. The great
portion of the gesture-only metaphors offers evidence clashes with the prediction of the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Lexical Semantic Hypothesis but agrees with the prediction of the Interface Hypothesis.
Generally, the present study on the cross-modal expressions of metaphors support the
Interface Hypothesis more.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
115
CHAPTER 6
CONCLUSION
An important claim of the Conceptual Metaphor Theory is that the locus of metaphor is
in thought. This view indicates that metaphors can be realized in different modalities.
Following this view, the thesis examined the linguistic and gestural manifestations of
conceptual metaphors in conversational discourse. Section 6.1 provides a summary of the
thesis. Section 6.2 presents the limitations and direction for future study.
6.1 Summary of the Thesis
The thesis aimed to explore people’s habitual cross-modal expression of metaphors in
daily conversation, and to investigate the collaboration of language and gesture in the
expression of metaphorical thoughts. The present study focused on the metaphors realized
concurrently in language and gesture, and the metaphors realized in gesture exclusively.
These two groups of metaphors were investigated with regard to metaphor types,
source-domain concepts, target-domain concepts, source-to-target correspondences, and
temporal patterning of speech and gesture. The present study also examined whether the two
groups of metaphors were similar or different with regard to the above issues.
According to the past research (Lakoff &Johnson 1980c, 1999; McNeill 1992; Talmy
1996), nine metaphor types were identified. In the current data seven of them were found:
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
causation metaphor, conduit metaphor, container metaphor, entity metaphor, fictive-motion
metaphor, orientation metaphor, and complex metaphor. Entity metaphor is prone to be
conveyed in metaphors in the gesture-only group. In both the language-gesture and
gesture-only groups, entity metaphor and orientation metaphor are the most common
metaphor types to be expressed in daily communication. Our experience of discrete objects
offers “a further basis...that goes beyond mere orientation” (Lakoff and Johnson 1980c: 25).
Understanding abstract concepts in terms of objects then allow us to project various
experiences of object to the concepts. Thus, it is likely that entity metaphor is frequently used
to conceive abstract concepts. Spatial orientations provide an “extraordinary rich basis” for
reasoning about other concepts, and orientation metaphor are closely relate to our physical
and cultural experiences (Lakoff and Johnson 1980c: 25). In some cases, it is difficult to find
an alternative to talk about a concept without the help of the orientation metaphor. Hence, the
orientation metaphor is often utilized to conceptualize abstract concepts.
Based on the notion of image schemas (Johnson 1987; Cienki 1997; Clausner & Croft
1999; Santibáñez 2002), nine kinds of source-domain concepts were distinguished in
previous studies. Seven of them were found in the current data: ACTIVITY, CONTAINER,
FICTIVE-MOTION, FORCE, OBJECT, PATH, and SPACE. Metaphors in the language-gesture
group tend to employ SPACE as their sources; metaphors in the gesture-only group are apt to
utilize OBJECT as their sources. Taken the two groups together, OBJECT is the most frequent
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
source for entity metaphors; SPACE and PATH is the second most frequent source for
orientation metaphors. Concerning the target-domain concepts, the present study focused on
the targets that had at least five tokens in the data. Eight kinds of targets were identified:
GROUP, MENTAL ACTIVITY, (physical) ACTIVITY, DEGREE, SEQUENCE, SPEECH CONTENT,
STATE, and TIME. SPEECH CONTENT is inclined to be conceived by metaphors in language
and gesture. Targets like STATE, PHYSICAL ACTIVITY, GROUP and MENTAL ACTIVITY are
prone to be conceptualized by metaphors in gesture exclusively. Among all the 247
metaphoric expressions, STATE, TIME, and PHYSICAL ACTIVITY are the common targets to
be manifested through metaphors.
With regard to the metaphorical mappings between the two domains, the present study
analyzed the one-source-to-many-targets correspondences and the many-sources-to-one target
correspondences. Excluding FORCE which has merely a token, all the sources could be used
to conceive multiple targets. OBJECT can be employed to conceptualize a great variety of
targets; concepts like STATE and TIME are the common targets mapping to OBJECT. SPACE is
usually employed to conceive concepts like TIME, SEQUENCE, and DEGREE. The sources
PATH, FICTIVE-MOTION, ACTIVITY,and CONTAINER are utilized to conceptualize multiple
targets as well. According to Lakoff and Johnson (1980b), metaphors have partial nature in
the way that a metaphor only profiles a certain aspect of a concept. An abstract concept is
likely to be conceptualized in terms of more than one concrete concept. In the current data,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
TIME, MENTAL ACTIVITY, SPEECH CONTENT, SEQUENCE, and DEGREE are the targets
which are conceived through multiple sources. These results demonstrate the common types
of metaphors, the source- and target-domain concepts, and the source-target correspondences
in daily conversations.
Furthermore, the collaboration of speech and gesture enables us to look at the
hypotheses of speech-gesture productions—the Lexical Semantic Hypothesis and the
Interface Hypothesis. The present study examined the two hypotheses from the following
aspects: the temporal patterning of speech and gesture, the linguistic unit to influence the
content of a metaphoric gesture, and the semantic coordination of speech and gesture in
conveying metaphors. According to the Lexical Semantic Hypothesis, gestures can help
lexical search when one has difficulty to retrieve a word in speech; hence, gesture strokes
may precede the associated speech. Since gestures are thought to be generated from the
semantics of the lexical items, it is predicted that gestures can only encode information that is
conveyed in accompanying speech and that the linguistic unit to determine the content of
gestures is a single word. However, the Interface Hypothesis suggests that gestural content is
shaped on-line by information which is exchanged between linguistic and spatio-motoric
thinking. Thus, gestures temporally synchronize with related speech, and the relevant unit to
influence the content of a gesture is a processing unit—which can be a clause to a certain
extent—for speech production. Since gestures are thought to be produced from an interface
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
between linguistic and spatio-motoric information, they may encode what is conveyed or not
conveyed in accompanying speech. Results from the current data show that many metaphoric
gestures synchronize with their associated speech, which supports the view of the Interface
Hypothesis more. Although most of the lexical affiliates of the metaphoric gestures are words,
a substantial portion has the gesture associated with a phrase rather than with a single word.
This portion of data provides evidence in contrast to the Lexical Semantic Hypothesis. In the
current data, over a half of the metaphoric expressions are gesture-only metaphors in which
language expresses the target-domain concepts and gesture expresses the sour-domain
concepts. Gesture-only metaphors reveal that the information encoded in speech and gesture
may differ from each other in conveying metaphors. This portion of data opposes to the
prediction of the Lexical Semantic Hypothesis but agrees with the Interface Hypothesis. In
general, results based on the current data support the Interface Hypothesis more than the
Lexical Semantic Hypothesis.
6.2 Limitations and Future Study
The investigation of the cross-modal expressions of metaphors can be extended in
future study to explore several issues which are not discussed in this thesis. The first issue is
how metaphors are embodied in daily experiences. Image schemas have been introduced to
the studies on metaphors (c.f., Johnson 1987; Lakoff 1987). Image schemas, the recurring
dynamic patterns of our sensory-motor experience, are seen as the primary sources of
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
metaphors. The present study also incorporates the notion of image schema to categorize the
source-domain concepts. Scholars proposed that some schemas are more general and cover
more specific schemas (Johnson 1987; Cienki 1997; Clausner & Croft 1999; Santibáñez
2002). Nevertheless, this thesis merely employs the image schemas at the general level (e.g.,
OBJECT, SPACE, and FORCE) to investigate the sources. According to past studies (Clausner
& Croft 1999; Santibáñez 2002), the image schemas in the specific level comprise more
& Croft 1999; Santibáñez 2002), the image schemas in the specific level comprise more