1. Introduction
1.2 Research Questions of the Study
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The opposite interpretation of being the removed entity (i.e. NPOUT) can be hardly accepted because new information is seldom introduced to be the replaced entity (NPOUT) in our live experience.
In addition, we hypothesized that if the NP is old information, it would show the neutral preference for either NPIN or NPOUT. This hypothesis was based on the
intuition that something discussed or mentioned in the prior context could be reasonable to be either NPIN or NPOUT.
The two hypotheses demonstrate the potential of old/new information to
distinguish the role of NP if both of them were attested in the natural language use. In particular, the first hypothesis that new information prefers NPIN may help distinguish the ambiguous role of the NP in the sentence patterns without the role-predicting preposition, e.g., [NPAGENT + Verb + NPIN/OUT].
Lastly, since linguistic expressions may reflect the variant construals (Langacker, 2008) of the event in conceptual representation, we will uncover the different
construals of the substitution event by examining the sentence patterns of SUBSTITUTE in natural-occurring language.
1.2 Research Questions of the Study
In this thesis, four research questions will be addressed for investigation:
1. What sentence patterns of SUBSTITUTE are displayed in natural language use?
2. What is the distribution of each sentence pattern in natural language use, especially the sentence patterns with and without the role-predicting prepositions?
3. What is the tendency between ‘old/new’ information in the information structure and the role of the NP in the sentence patterns of SUBSTITUTE?
4. What construals does SUBSTITUTE possess in the event of substitute?
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The present thesis is structured as follows. In Chapter Two, we will review the related literature of (a) construal and its relation to linguistic expressions in
conceptual approach and (b) the impact of the information structure on sentence patterns in discourse. In Chapter Three, we will present the methodology, specifically the annotation of the participants in the event of substitution and the information status of the NPs in sentence patterns. In Chapter Four and Five, we will present the results of the conceptual analysis of the sentence patterns and the relation between information status and sentence patterns, respectively. Lastly, in Chapter Six we will conclude the findings of the thesis by answering the four research questions. In addition, according to the findings, we will propose the pedagogical implications and indicate the limitations of the thesis.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 2
LITERATURE REVIEW
To tackle the issue pertaining to the configuration of grammatical structure, we review the linguistic theories explaining the configuration of simple clause structure, defined as “a sentence with one main clause” in Carter & McCarthy (2006:488).
The chapter is structured as follows. In section 2.1, we will introduce the conceptual approach to simple clause structure in Langacker (1991, 1999, 2008).
Fillmore’s (1982, 1985, 2006) Frame Semantics will be introduced as well. Then, in section 2.2, the scope is extended beyond the boundary of clauses and information structure will be reviewed. In section 2.3, we will summarize this chapter.
2.1 Conceptual Approach to Clause Structures
In this section, we present two notions advocating that humans describe their experience of events by means of conceptual structure. First, we introduce the notion of ‘Frame Semantics’ (Fillmore, 1982, 1985, 2006) as the foundation of conceptual approach in section 2.1.1. Then, on the ground of ‘Frame Semantics’, we review Langacker’s (1991, 1999, 2008) proposals of ‘construal’ in section 2.1.2 and
demonstrate the influence of ‘construal’ in the encoding of clause structure in section 2.1.3. Lastly, in section 2.1.4 we will present the approach adopted in this thesis.
2.1.1 Frame Semantics
The theory of Frame Semantics was proposed to respond the criticisms and limitations of semantic roles (Fillmore, 1982, cited in Fillmore, 2006). Specifically, Fillmore acknowledged that semantic roles fail to offer detailed semantic description:
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
This theory of semantic roles fell short of providing the detail needed for semantic description; it came more and more to seem that another independent level of role structure was needed for the semantic description of verbs in particular limited domains.
(Fillmore, 1982:377)
Fillmore (1982:115) proposed the abstract ‘scenes’ as the basis in Frame Semantics (‘scene’ was later replaced by ‘frame’ in Fillmore (1985), we use ‘frame’
hereafter). As explained in Fillmore and Baker (2010:314), Frame Semantics is “the study of how, as a part of our knowledge of the language, we associate linguistic forms (words, fixed phrases, grammatical patterns) with the cognitive structures—the frames” (p. 314). In other words, Frame Semantics advocated the continuities between language and our experience or encyclopedic knowledge in conceptual structure.4 The ‘frame’ is the schematic and conceptual representation which entails the necessary properties, experiences, and knowledge as the basis and premise to understand the semantic structures and meanings of the verb. That is, frames are the presupposed knowledge and experience to understand the meanings of a word. For example, the meanings of a set of semantically related verbs, including buy, sell, spend, etc. can be understood by knowing what takes place in a commercial transaction in the {COMMERCIAL TRANSACTION}frame(Fillmore & Atkins, 1992:78).5
The {COMMERCIAL TRANSACTION}frame is the knowledge base comprising our experiences in a commercial transaction event, for example, buying books in a
bookstore. Progressively, the experience of commercial transaction events establishes
4 In this thesis, we used ‘conceptual structure’ and ‘cognitive structure’ interchangeably.
5 Note that the name of frames is presented in lower capital embedded in curly brackets.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the required knowledge and concepts in the {COMMERCIAL TRANSACTION}frame. For instance, as indicated by Fillmore and Atkins (1992:78), there is a ‘BUYER’
purchasing ‘GOODS’ and a ‘SELLER’ selling the ‘GOODS’ in exchange of the ‘MONEY’ from the ‘BUYER’.6 These four elements were termed the ‘frame elements’ (FEs) (Fillmore & Baker, 2010:321) of the {COMMERCIAL TRANSACTION}frame. Since the lack of each element may cause the commercial transaction event incomplete (the
‘BUYER’ cannot buy the ‘GOODS’ without paying ‘MONEY’), these elements are the
‘core FEs’ of the {COMMERCIAL TRANSACTION}frame. In contrast, the ‘TIME’ in which the commercial transaction event occurs is negligible in the frame, and thus
‘Time’ is the non-core (or peripheral) FE of the {COMMERCIAL TRANSACTION}frame.
In a different point of view, Petruck (1996) indicated that “the words, that is, the linguistic material, evoke the frame (in the mind of a speaker or hearer); the
interpreter (of an utterance or a text in which the words occur) invokes the frame”
(Petruck, 1996:1). In other words, the verbs, e.g. buy, sell, and spend, are the linguistic prompts which activate our knowledge base of the {COMMERCIAL
TRANSACTION}frame to encode or interpret the event of commercial transaction.
It is important to clarify that the requirement of core FEs in a frame does not mean their realization in a sentence is required as well. While all the core FEs could be realized (see 2.13a), oftentimes not every core FE is realized, as in (2.1b).7
(2.1) a. [BUYER Carla] bought [GOODS the computer] [SELLER from Sally]
[MONEY for $100]. (Petruck, 1996:3)
b. [SELLER Sally] sold [GOODS the computer] yesterday.
6 In this thesis, the frame elements are presented in lower capitals within single quotations, for example, ‘BUYER’, in order to be distinguished from the semantic roles (e.g., ‘Agent’).
7 In this study, we only annotated the core FEs in the sentence. Since yesterday refers to the non-core FE ‘Time’, we did not label its frame element in (2.13b).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Therefore, the core FEs are necessary in the sense that none of the core FEs could be deleted in the knowledge base of the frame. For example, the {COMMERCIAL
TRANSACTION}frame would be incomplete if one of the core-FEs (e.g., ‘SELLER’) is not part of the frame.
Another important concept in Frame Semantics is ‘perspective’. Fillmore and Baker (2010) proposed that “different lexical items (e.g., buy and sell) evoke frames with different perspectives on an abstract event (commercial transaction)” (p. 330).
Consider (2.1) again. On the one hand, in (2.1a) the verb buy takes the perspective of the ‘BUYER’, so that the ‘BUYER’ is realized as the subject of buy. On the other hand, in (2.1b), the verb sell takes the perspective of the ‘SELLER’, and thus the ‘SELLER’ is realized as the subject of sell. While (2.1a) and (2.1b) describe the same event that
‘Sandy sold the computer to Carla’, the two verbs take different perspectives on the event, and thus give rise to different realizations of core FEs in the sentences.
Using Frame Semantics as the basis, the Berkeley FrameNet project has
developed an online lexical resource with corpus evidence. The aim of the project is
“to document the range of semantic and syntactic combinatory possibilities of each word in each of its senses” (Ruppenhofer et al., 2016:7). In other words, based on the frames evoked by a word (usually a verb), the FrameNet aims to capture how the frame elements can be realized in the syntactic structure of the verb in natural occurring language. For example, in the verb sell, the ‘SELLER’ is realized as the subject while the ‘GOODS’ is realized as the direct object (see 2.1b). At the same time, the verb sell can also be expressed in Sally sold Carla a computer where the direct object Carla is the ‘BUYER’ rather than the ‘GOODS’, and the ‘GOODS’ (a computer) is realized in the indirect object. These combinatory possibilities of the verb sell have been documented as such in the lexical source of the FrameNet.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Then, we consulted the FrameNet for the frames presupposed by SUBSTITUTE (https://framenet.icsi.berkeley.edu/fndrupal/framenet_search). SUBSTITUTE, as indicated in the FrameNet, evokes two different frames as its conceptual knowledge base: one is the {REPLACING} frame and the other is the {TAKE PLACE OF} frame. The definition of each frame and its core FEs are provided in Table 2.1 and Table 2.2.
In the {REPLACING} frame, ‘AGENT’, ‘NEW’, and ‘OLD’ were identified as the core FEs. The definition of this frame suggests the knowledge and experience that an
‘AGENT’ uses the ‘NEW’ to take the place of the ‘OLD’.
Table 2.1 The definitions of the {REPLACING}frame and its core FEs
(https://framenet2.icsi.berkeley.edu/fnReports/data/frameIndex.xml?frame=Replacing) Frame REPLACING
Definition An ‘AGENT’ changes the filler of a role by placing an ‘NEW’ filler in the position after the ‘OLD’ filler ceases to occupy the position.
Core FEs ‘AGENT’ The conscious entity, generally a person, that performs actions resulting in the ‘NEW’ entity occupying the position.
‘NEW’ The ‘NEW’ entity is the person or thing that the ‘AGENT’ sets to fill a role
‘OLD’ The entity that formerly occupied the position.
The {REPLACING} frame could be exemplified as in (2.2). The ‘AGENT’ Wilkinson could be a coach or a manager who instigates the action of substituting by introducing the ‘IN’ (David R.) as the new player to take the place of the ‘OUT’ (Frank Strandli).
As for with 8 minutes remaining, it is not annotated with a frame element in this thesis due to its status as a non-core frame element, ‘Time’.
(2.2) [AGENT Wilkinson] substituted [NEWFrank Strandli] [OLD with David R.] with 8 minutes remaining. (FrameNet)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
According to FrameNet, the other frame evoked by SUBSTITUTE is the {TAKE PLACE OF} frame, in which three core FEs are displayed in Table 2.2. The first two are the ‘NEW’ and the ‘OLD’, and the third core FE could be either ‘FUNCTION’ or ‘ROLE’.
In general, this frame describes the knowledge and experience that a ‘NEW’ takes the place of an ‘OLD’ by occupying the former ‘ROLE’ or ‘FUNCTION’ of the ‘OLD’.
Table 2.2 The definitions of the {TAKE PLACE OF}frame and its core FEs
(https://framenet2.icsi.berkeley.edu/fnReports/data/frameIndex.xml?frame=Take_place_of) Frame TAKE PLACE OF
Definition A ‘NEW’ filler occupies a ‘ROLE’ or serves a ‘FUNCTION’ after the
‘OLD’ filler ceases to occupy the position. In many cases, the ‘ROLE’ or
‘FUNCTION’ is implicit.
Core FEs ‘NEW’ The new filler of the ‘ROLE’ that was previously occupied by the ‘OLD’ filler.
‘OLD’ The entity that occupied the ‘ROLE’ or served the
‘FUNCTION’ before the ‘IN’ filler.
‘FUNCTION’ The (generally desirable) state of affairs that the ‘NEW’ and ‘OLD’ entities are involved in bringing about.
‘ROLE’ The category that the ‘NEW’ and ‘OLD’ entities fit into as a result of the ‘FUNCTION’ they serve.
The {TAKE PLACE OF} frame could be exemplified as in (2.3).
(2.3) [NEWCharcoal] can also substitute [OLD for fossil fuels], which in some places is an urgent need. (FrameNet)
In (2.3), Charcoal is the ‘NEW’ realized as the subject of SUBSTITUTE and takes the place of the ‘OLD’ which is realized as the oblique, for fossil fuels. As for the third core-FE, either ‘ROLE’ or ‘FUNCTION’,is absent in (2.3).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
This section has presented the point of view of Frame Semantics which emphasizes that linguistic expressions relate to the evoked knowledge structure or experience of the words (i.e. frames) and that different linguistic expressions are the results of the realizations of the frame elements in the frames. For here, the
realizations of the core FEs of SUBSTITUTE was exemplified to illustrate this
notion. In general, Frame Semantics hold the tenet that the use of language reflects the experience and knowledge embedded in the frame underlying the words.
In the next section, we will introduce Langacker’s (1991, 1999, 2008) notions that language is the conceptual symbolization providing different devices for us to encode the experience of events.
2.1.2 Construal in Cognitive Grammar
On the ground of cognitive approach, Langacker (1991, 1999, 2008) developed the linguistic theory known as Cognitive Grammar. The basic tenet of Cognitive Grammar is that language is “symbolic in nature” (Langacker, 2008:5). To be more specific, language is structured by symbols, that is, the symbolic assemblies which are the form-meaning pair in nature (Langacker, 2008). For example, a simple lexical item, for example, cat, is symbolic in that its alphabetically structured form (cat) or its phonological structure /kæ t/ is paired with its meaning. Furthermore, even the
grammar of a language, known as the system which combines elements into complex expressions, is considered symbolic. As indicated by Langacker (2008), “lexicon and grammar form a gradation consisting solely in assemblies of symbolic structures” (p.
5). For instance, the grammatical categories (e.g., ‘nouns’) are symbolic because they carry the more schematic meaning of ‘things’. In this regard, Cognitive Grammar rejects the ‘autonomous’ nature of syntax, that is, the separation of syntax and semantics, in Generative Grammars (Chomsky, 1986).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
As indicated above, a symbolic assembly contains a form-meaning pairing.
Then, a question arises: where does the meaning come from? Evans & Green (2006) explained that “[t]he meaning associated with a linguistic symbol is linked to a particular mental representation termed a concept. Concepts, in turn, derive from percepts” (p. 7). Consider the meaning of cat again. We interact with cats in the physical world through our sensory perceptions, such as vision and touch. Through vision, we know the appearances and colors of cats, and through touch we feel the fluffy furs of cats. All these are our ‘percepts’ of cats. Then, progressively, these percepts of cats are integrated into the mental (or conceptual) representation, which, in turn, forms the ‘concepts’ of cats. Imagine a scenario in which you are directing your friend’s attention to a cute cat on the street by Look at the cat! In your utterance the cat derives from your concept of cat in the conceptual representation. This
scenario demonstrates that the meaning of the symbolic assembly cat comes from the conceptual representation of cat rather than any specific cat in the physical world.
With this understanding, Langacker proposed that “a meaning consists of both conceptual content and a particular way of construing that content” (p. 43). While the
‘conceptual content’ refers to, for example, the concept of cat in the conceptual representation, the term ‘construal’ is defined as “our manifest ability to conceive or portray the same situation in alternate ways” (p. 43).
Langacker adduced a half-full glass of water for illustration (see Figure 2.1). At the conceptual level, the conceptual content is in the “neutral manner” without any imposed construals on it, as the left-most diagram in Figure 2.1. Then, as we encode the conceptual content into language, inevitably, different construals may be imposed on the conceptual content. In (2.4), Langacker displayed the encoded sentences corresponding to the four possible construals in Figure 2.1.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Figure 2.1 Construals in conceptual content (adapted from Langacker, 2008:44)
(2.4) a. the glass with water in it (Langacker, 2008:43)
b. the water in the glass (ibid)
c. the glass is half-full (ibid)
d. the glass is half-empty (ibid)
Sentence (2.4a) represents the first construal which designates the container, the glass;
(2.4b) takes the second construal designating the liquid, the water, in the container;
(2.4c) designates the relationship between the container (the glass) and the half-filled water; (2.4d) designates the relationship between the container and the half-void space in the container. Sentences in (2.4) demonstrate how the identical conceptual content can be construed differently in language from (2.4a) to (2.4d), respectively.
As demonstrated in (2.4), different locus of the conceptual content could be selected in these four construals. In (2.4a), the container is selected; in (2.4b), the liquid is selected; in (2.4c), the relationship between the container and the liquid is selected; in (2.4d), the relationship between the container and the void is selected. The selection of locus of the conceptual content, as proposed by Langacker (2008),
reflects the mechanism of “prominence” (p. 66) in construal. In this thesis, among others, we concentrate on two kinds of prominence in language structure: the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‘profiling’ and the ‘trajector-landmark alignment’, the two main notions used in this thesis later.
2.1.2.1 Profiling
Within the conceptual structure, the conceptual content, as the glass of water in Figure 2.1, is considered the conceptual ‘base’ as “the immediate scope in active domains” (Langacker, 2008:66). The proposal of the conceptual ‘base’ is akin to the
‘frame’ in the Frame Semantics, that is, the encyclopedic knowledge or experience.
For ‘profiling’, it refers to the “the specific focus of attention” (Langacker, 1999:7) in the conceptual ‘base’. Taken together, ‘profiling’ is defined in Langacker (2008) as follows.
Within this onstage region, attention is directed to a particular substructure, called the profile. Thus an expression’s profile stands out as the specific focus of attention within its immediate scope.
(Langacker, 2008:66)
In the example of the half-full glass of water, the integrity of the whole conceptual content, including the container, water, and the void, is the conceptual ‘base’ for selection. Then, the selected locus (i.e. the particular substructure) of the ‘base’ is the
‘profile’ of a certain construal. For instance, in the glass with water in it, the ‘profile’
is the container encoded by the noun phrase, the glass; then, the whole conceptual content in Figure 2.1 is the ‘base’ of the ‘profile’.
Furthermore, in the same base, the profile could be either a ‘thing’ or a
‘relationship’ (Langacker, 2008:67). In the glass with water, the expression profiles
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the ‘thing’ in the glass.8 In contrast, the preposition with profiles the ‘relationship’
between the glass and the water. The fundamental distinction between the ‘thing’ and the ‘relationship’, as indicated by Langacker (1991), lies in whether the conception can be conceptualized independently. While the ‘thing’ can be conceptualized independently, the ‘relationship’ “does not exist independently of its participants”
(Langacker, 1991:14). For example, in John broke the glass, we can independently form the conceptions of ‘thing’, such as John and the glass. However, we cannot conceptualize the verb break without referring to its participants, John and the glass.
Therefore, Langacker argued that the profiling of a ‘thing’ is “conceptually autonomous” (p. 14) and that a ‘relationship’ is “conceptually dependent” (p. 14).
On top of that, Langacker (2008:369) applied profiling to the force-dynamic event of glass breaking, as in Floyd broke the glass with a hammer. As a mechanism directing attention to a certain portion of the conceptual base, Langacker (2008:369) argued that different linguistic expressions demonstrate the varying profiled portion of the event. For instance, the sentence A hammer broke the glass shows that the
attention directed to the interaction between the hammer and the glass, and another expression The glass broke only focuses on the result of the glass. These three expressions show that language allows us to focus on different portion of an event, and ‘profiling’ is the cognitive mechanism underlying these expressions.
This sub-section illustrates the notion of profiling which represents the
This sub-section illustrates the notion of profiling which represents the