1.4 Objectives
2.1.4 Summary and conclusion
國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
different positions, it is less feasible for broadcast news.
2.1.4 Summary and Conclusion
From the previous studies, we have learned all the three approaches agree upon
that broadcast news is composed of news kernel (i.e., the opening anchor sequence)
and news body (i.e., the story proper). However, they have not yet reached agreement
on the structural elements of news body. Researchers of narrative approach (Hartly,
1982; Fiske, 1987; Graddol, 1994; Thompson, 2005) treat broadcast news as a variant
of narrative, which can be produced through multiple modes together like verbal and
visual modes. News kernel is equal to the first element—abstract in narrative. The
elements in the news body, however, differ from those of narrative. To attract the
audiences’ attention, news body sometimes begins with something visually attractive
(Thompson, 2005; Lanson and Stephens, 1994; Chang, 2007). Then the whole news
story is developed along with orientation, complicating actions, resolution and coda.
News-as-commentary approach (Montgomery, 2007) is valuable in that it points
out a commentary relationship between both words to images and shot boundaries to
clause/sentence boundaries. Thus, we may infer that shot shifts as the events or topics
are switched. Though emphasizing the commentary relationship between the two
modes, news-as-commentary approach does not touch how events are arranged in a
news item. Thus, the approach may only be used when we examine the correlation
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
between shot shift and topic shift.
Satellite approach (Thornborrow and Fitzgerald, 2004), opposing the tradition of
viewing news as a form of narrative, sees news report as a “thickening” relationship
between news kernel and news body. It proposes that each section/paragraph of news
kernel can be viewed as an intra-linked story, which encompasses category, action and
reason. Then, the three elements of the intra-linked story recur in the following news
body. Indeed, news body do serve the function of “thickening” the elements in news
body, since news kernel is a virtually a brief version of news body. Yet, based on our
observation, each intra-linked stories defined by Thornborrow and Fitzgerald (2004) is
still connected to each other according to their temporal sequence in the real world.
For instance, in example (1) in 2.1.3, the actions initiated by habitants near Galicia
(i.e., the first strand) took place prior to the Portuguese government’s turn-down of the
request to dock near Portugal.
After comparing the three approaches, we have learned that each approach has its
pros and cons. We will adopt narrative approaches to serve as a framework of the
outlook of a piece of news and use news-as-commentary approaches to examine the
co-relationship between shot shift and topic shift.
2.2 Marker of Discourse Structure
Any type of discourse can be segmented into smaller sections, each of which is
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
semantically coherent individually (Oliverira, 2000). The propositions included in one
discourse segment can be subsumed under a macro-proposition or summarized in a
single word or phrasal title (van Dijk, 1988; McCarthy, 2002). They can be either
termed as topics (van Dijk, 1988) or thematic paragraphs (Givon, 1983). The boundary
of each topic or thematic paragraph usually indicates the change of time/period, place,
action, participants, or possible world. In other words, the introduction of a new topic
is usually marked in a “semantically conspicuous way,” which is usually realized in
forms, intonation patterns or pausing (van Dijk and Kintsch, 1983).
In the present study, we will only review studies on forms. Previous studies
(Schiffrin, 1987; Fraser, 1999; Li, 1985; Yu, 1990; Chen, 1990; Qin, 1998; Miracle,
1991; Chui, 2002; Wang, 2001, 2005) on boundary markers are fruitful. The genres
cover written narratives, oral narratives and daily conversation. Their results show that
there exists a correlation between boundary markers and the conspicuousness of
discourse units. Thus, in the following section, we will introduce the categorization of
discourse markers first and then the relationship between discourse units of different
size and their corresponding boundary markers.
To begin with, discourse markers, according to Schiffrin (1987), are cohesive
devices which can not only indicate the connections between two sequential
propositions but also bracket units of talk, which may refer to speech acts, utterances,
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
turns or sentences. They have been extensively investigated since the 1980s. The
similar terminologies include cue phrases, clue words, discourse connectives and
discourse particles (Hirschberg & Litmaw, 1993). Most of the previous studies agree
that discourse markers are able to mark discourse structures. For example, the use of
“but” at the beginning of turns in conversation may indicate topic or sub-topic shift
(McCarthy, 2000). Therefore, their occurrences can help the audiences comprehending
the content by indicating the shift of topic (Schiffrin, 1987).
As to the subcategories of discourse markers, Schiffrin (1987) has divided them
into four kinds of lexical expressions—conjunction, interjections, adverbs and lexical
phrases. Similarly, Fraser (1999) also asserts that discourses markers can belong to
different syntactic classes, such as connectives, adverbs, and prepositional phrases. In
terms of positions, discourse markers can be further divided into two types. Some
occur at the beginning of a discourse unit yet some at the end. At the semantic level,
they help the narrators establish coherence; at the discourse level, they segment
discourse into several components. Therefore, the cohesive devices can not only
denote coherence but also function as markers indicating the demarcation of units.
With regard to Mandarin Chinese, markers used to signal boundaries include
connectives, anaphoric variants, topic shift fillers and sentence-final particles. In
addition to these markers, we suggest adding shot shift, since visual image plays an
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
important role in broadcast news. We will introduce these boundary markers in the
order above. First, concerning connectives, Yu (1990) has conducted a thorough study
on the discourse function of connectives as well as the correlation of connectives and
anaphora in written narratives. She discovers that four types of connectives—additives,
adversatives, causal-effects and temporals are able to mark discourse segments of
different levels. Simply stated, different types of connectives tend to occur at different
unit boundaries. Additives usually connect micro-structures which are related to one
topic. Adversatives, however, frequently connect two topics each of which has its own
topic. Therefore, the presence of adversatives may denote the topic is shifted.
Causal-effects can function as global markers and local markers. Similarly, temporals
can also indicate macro-structure as well as micro-structure.
Chen (1990) compares the cohesive devices in spoken and written narrative. She
finds that contrajunctions such as raner 然而 ‘yet,’ bugo 不過 ‘but,’ and
meixiangdao 沒想到 ‘unexpectedly’ can serve to signal a new topic. Besides,
contrajunctive expressions which indicate the meaning of opposition such as
shishisang 事實上 ‘in fact,’ qishi 其實 ‘in fact,’ and xiangfande 相反的 ‘in
contrast’ are also able to start a new topic.
Qin (1998) analyzes the narrative structures in spoken Mandarin and finds some
correlations between the discourse markers and the six components of a narrative. In
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
her data, the utterance-initial particle na 那 is the most prevalent discourse marker,
which is able to mark orientation, complicating action, and evaluation. Discourse
markers such as houlai 後來 ‘then,’ keshi 可是 ‘but,’ and yinwei 因為 ‘because’
often precede orientation. The successive events are usually connected by ranhou 然
後 ‘and then,’ houlai 後來 ‘then,’ and jieguo 結果 ‘as a result.’ Evaluation often
begins with markers such as jieguo 結果 ‘as a result,’ houlai 後來 ‘then’ and suoyi
所以 ‘so.’ As to coda, it is predominately marked by suoyi 所以 ‘so.’
In addition to connectives, anaphora is also a good indicator of topic shift.
According to Givon (1983)’s Iconicity principle, if a topic or referent is discontinuous
or less accessible, a referential form with more coding materials is chosen to make the
topic more informative. On the other hand, if a topic or referent is continuous or
accessible, a referential form with less coding materials will be used. With regard to
Mandarin Chinese, Li (1985, as cited in Yu, 1990) has also done a thorough study on
the three variants of anaphora. According to her categorization, full NPs, pronouns and
zero anaphora are the morphological variants of anaphora and they are related to
discourse units at different levels. Full NPs tend to occur at the beginning of a larger
unit and can progress the story line. Pronouns mark the beginning of a smaller unit like
a topic chain with minor break in continuity. Zero is an internal tie which connects two
closely related clauses. Therefore, the presence of full NPs and pronouns indicates the
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
topic is switched.
The third category of boundary marker is topic shift fillers, including hao 好
‘okay,’ er 而 ‘and/but,’ name 那麼 ‘then,’ and zhege 這個 ‘this.’ Previous studies
on hao 好 ‘okay’ (Miracle, 1991; Wang, 2001, 2005) examine both its literal
meanings and discourse functions. The researchers agree that in addition to expressing
the meaning of “good,” “fine,” “nice,” and “okay,” hao 好 can serve as a discourse
marker. It signals the closure of the previous unit and the transition to the new topics at
the textual level of discourse (Wang, 2005). As to er 而 ‘and/but,’ name 那麼 ‘then,’
and zhege 這個 ‘this,’ they have similar functions. Er, 而 ‘and/but’ can denote
additive, adversative, and causal-conditional. Besides, it is able to serve as a topic shift
point (Chen, 1990; Hong, 2008). These topic shift fillers are often viewed as a
mechanism for the speakers to earn more processing time as the topics shift. Thus, the
presence of topic shift fillers is a possible indicator of topic changes.
The fourth boundary marker is sentence-final particle, which is also termed as yu
qi chi 語氣詞 ‘mood words’, including la, 啦, ne 呢, o 喔, a 啊, ma 嗎, ba 吧, etc.
They mainly occur in informal context and function as expressing speakers’ emotions
and attitudes (Li, 1999). Traditional research often lists the meanings of sentence-final
particles by offering examples (Diao, 2000; Tsai, 2002, cited in Wu, 2006). However,
in a thorough examination of sentence-final particles, many researchers (Li, 1999; Hu,
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
1993, cited in Wu, 2006; Wu, 2006) point out that the inferred meanings of
sentence-final particles are actually a combination of the propositions of the whole
sentence, intonation and the context. Li (1999) further suggests that the meaning of
sentence-final particles is general. The specific meanings are derived in different
context7. The indexical function suggests that sentential-final particles should be
treated as discourse markers. For example, la 啦 in Taiwanese usually occurs at the
end of a discourse unit of a larger level, such as the changes of topics.
The last type of boundary marker is shot shift. According to Montgomery (2007),
there is nearly a 50% correspondence between shot boundaries and clause/sentence
boundaries. The results may infer that there is a correlation between shot shift and
discourse segments because the propositions coded in clauses/sentences are matched
with relevant scenes. Thus, it is reasonable to assume that shot shift goes with the shift
of topics and accordingly serves as indicators of topic switching.
Based on the previous studies, we have learned that sentence-final particles,
connectives, anaphoric variants, topic shift filler and shot shift are able to indicate the
boundary of a discourse unit. Sentence-final particles often occur in informal context
or conversation. Accordingly, they are less likely to be found in a formal genre like
7 Li’s (1999) analysis on la 啦 in Taiwanese provides us a good example. She mentions that la 啦 in 緊吃啦 denotes the meaning of “impatient” and “friendly” in the contexts where a mother talking to her son who is going to be late for school and where a host talking to his or her guest, respectively.
Therefore, la 啦 has a general meaning. The specific meanings, such as “impatient” and “friendly” will
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
broadcast news. However, we found only a few sentence-final particles, including la
啦, o 喔 and ne 呢 in our data, most of which are in news kernel and produced by
some anchors. Thus, we will exclude sentence-final particles in the present study.
Connectives, which help topic shifts more smoothly, are good indicators of discourse
units in narratives. They are also commonly seen in broadcast news, which usually
covers a series of events happening at different times, locations and involving different
persons. Besides, anaphoric variants, which are able to signal the discourse units of
different level, are as well useful indicators, because the focus of broadcast news is
human beings. Topic shift fillers are very likely to occur, especially in news kernel.
Shot shift, the distinctive feature of broadcast news, is also used as an indicator. Thus,
in the present study, we will use connectives, anaphoric variants, topic shift fillers, and
shot shift as the boundary markers and see their relationship with discourse units.
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
CHAPTER THREE METHODOLGY
In this chapter, data collection and description, transcription conventions and
analytical framework will be presented. Section 3.1 presents the data collection and
description. Section 3.2 presents the transcription convention. Section 3.3 introduces
the framework concerning the structure of a broadcast news item as well as the
potential boundary markers of each structure. Section 3.4 offers an example to
illustrate how the model operates.
3.1 Data Collection and Description
The data presented in this thesis were twenty-five pieces of news, recorded from
television news programs in Taiwan8. The channels were evenly distributed to reduce
idiosyncrasy, since each news organization may set its own prescriptions. Besides, the
genders of the news presenters, eight females and seven males, are also evenly
distributed in avoidance of gender differences. In regard to the subject matter, only
damage and criminal news are taken into consideration in that they occupy the highest
percentage (i.e., one-fifth) of a broadcast news section in Taiwan (Niu, 2006). Besides,
they are more akin to personal oral narratives, as defined by Labov and Waltezky
8 The channels include: Era-News (年代), ET-Today (東森), CTI (中天), FTV (民視), SET (三立), TVBS, TTV (台視), CTV (中視) and CTS (華視). The previous six channels are 24-hour television news channels and the last three are cable TV channels. The inclusion of the previous six channels is because they are professional news channels (Cai, 2006) and are probably the main source from which
‧
(1975) in that both of them cover a series of events, resulting in some reportable
damages or casualties. Their background information is given in Table 1.
Table 1 The information of the twenty-five pieces of news No. Data types Channel Reporting date
The topics of the 25 pieces of news include murder, suicide, accident, car accident,
robbery, abduction, fire, fraud, arson, shipwreck, arrest, harm, explosions and illegal
launching to make our data more diverse. Each topic was originally designed to
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
contain two or more pieces of news, but due to the limited time, the last seven topics
only contain one piece of news, respectively. Besides, each channel donated two or
three news items. The recording time is aimed at morning, noon, evening and midnight.
As for the length of a piece of news, each piece of news is one min and 37 sec on
average. News kernels are 24 sec in length and news report 73 sec in length, which is
three times the length of news kernel.
3.2 Transcription Conventions
TV news, though written in advance, is presented in spoken form. It is tactically
designed to sound as natural as speech (Cotter, 2001). Therefore, it exhibits the
characteristics of ordinary speech, such as pauses, rising, falling and level intonations.
We adopted Du Bois et al.’s (1993) transcription conventions for ordinary speech.
Under the transcription conventions, “speech is segmented into intonation units (IU),”
defined as “a stretch of speech under a single coherent intonation contour.” Its
boundary is commonly marked by four features: 1) pitch reset: the pitch value of a
new prosodic unit is usually higher than that of the previous one; 2) pause: an IU
usually ends with a longer pause; 3) lengthening: the final syllable of an IU is usually
lengthened; and 4) speech rate alternation: the speech rate is slow at the end of an IU
but is fast at the beginning of the next IU. In the present study, the acoustic features to
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
from acoustic features, because shot shift is considered to be related to event shift
(Montgomery, 2007), we also mark it in our text. All of the transcription symbols are
listed in Table 2.
Table 2 Transcription symbols
Features Symbols
Pause9 1) Long: …(N)
2) Short: ..
Intonation 1) Rising pitch: /
2) Falling pitch: \ 3) Level pitch:_
Shot shift >
3.3 Analytical Framework
Concerning the discourse structure of TV news, we chose the narrative approach
as our framework for two main reasons: 1) TV news shares similar structural elements
with narratives, and 2) these elements are presented in a canonical way as oral
narratives. Among the narrative approaches, we mainly followed Montgomery’s (2007)
and Labov and Waltezky’s (1975) models. To begin with, we took Montgomery’s
(2007) observations on the overall structure of broadcast news as our model, in which
a broadcast news item is composed of an obligatory news kernel and a succeeding
news body. A news kernel is a summary of the whole news story (“in a nutshell”). The
news body, on the other hand, clarifies and adds details to the news story. Nevertheless,
9 In Du Bois et al.’s (1993) framework, pauses are categorized into long, medium and short. Long pauses are beyond 0.6 seconds, medium ones are between 0.6 and 0.3 seconds, and short ones are below 0.3 seconds. However, in broadcast news, which is so constricted in time span, long pauses are seldom found. Thus, we divide pauses into only long and short ones. Long pauses are beyond 0.3 seconds and
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Montgomery’s study does not elaborate on the components of the two macrostructures.
To solve this problem, we adopted Labov and Waltezky’s (1975) analysis of narratives.
In their framework, a narrative consists of abstract, orientation, complicating actions,
evaluation, resolution and coda. These elements are also found in news. Yet some of
the terms and definitions are modified. First, the abstract in Labov and Waltezky’s
(1975) framework is roughly equal to the opening phrases in a news kernel. Like the
abstract, the opening phrases in a news kernel serve as a prelude to the upcoming news.
Its major function is to attract the audience’s interest in the news story. Orientation is
usually combined with the first event of the complicating actions, so we take it as
“oriented event 1.” Complicating actions are concerned with what happens in the news,
so we replace it with “main news events.” Resolution is replaced by “follow-ups,”
since news is an on-going story and often presents the latest development instead of
the final result, as the event has usually not yet ended. Evaluation remains the same as
it is in narratives. The types of coda differ greatly between broadcast news and
narratives. In narratives, codas are usually a few words or sentences concluding the
whole event or series of events. In broadcast news, however, news items always end
with a routine formula covering channel names, journalists’ identities and reporting
locations. Thus, we replace the term coda with “routine ending.” Besides the elements
abovementioned, after analyzing the data, we added one news-specific category in the
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
present study, the event scene presentation in the news body. The unit is characterized
as presenting newsworthy or shocking scenes and is always placed at the beginning of
as presenting newsworthy or shocking scenes and is always placed at the beginning of