2. Literature Review
2.3 Summary of the Chapter
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
analysis in the discourse level should be included in the investigation of clause structure.
2.3 Summary of the Chapter
In this chapter, we have reviewed theories and approaches relating to the configuration of clause structures. In section 2.1, we started with the conceptual approach toward clause structures, specifically Fillmore’s (1982, 1985, 2006) Frame Semantics and Langacker’s (1991, 1999, 2008) notions of construals, including
‘profiling’, ‘trajector-landmark alignment’, and ‘action chain model’. Then, in section 2.2, our review went beyond the sentence level and reviewed the information structure at the discourse level. Specifically, Prince (1992) provided a fine-grained taxonomy of information status, and Du Bois (2003) demonstrated the Preferred Argument
Structure in which information structure came into play in the configuration of constructions. The review in section 2.1 and 2.2 has demonstrated that both the construal underlying linguistic expressions and the information structure could be the contributing factors of the configuration of constructions. In other words, by
analyzing the construal and information structure in the sentence patterns of
SUBSTITUTE, we could reveal how language users construe the event of substituting and how they arrange the information structure of the sentence patterns. In the next chapter, we move on to introduce our methodology to analyze these two aspects of the sentence patterns of SUBSTITUTE.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 3 METHODOLOGY
This study aims to investigate the use of SUBSTITUTE as a verb in the
following aspects: (a) the construal of each sentence pattern, particularly the trajector-landmark alignment in profiled participants; (b) the distribution of each sentence pattern in natural occurring language; (c) the information status of the profiled participants in sentence patterns. In this chapter, we introduce the methodology we took to address these issues.
3.1 The Corpus
In order to investigate SUBSTITUTE in natural occurring language, we used the British National Corpus (BNC) as the source to retrieve the language data. It
possesses a 100 million word collection including written and spoken language. In written data, BNC covers newspapers, academic books, and other written texts. Its spoken collection ranges from business to government meetings in different contexts.
3.2 The Method for Extracting Data
In this study, we used ‘{substitute}_V*’ as the query term of the BNCweb. The result of the search returned 1314 hits in 659 different texts with the frequency of 13.37 instances per million words.
After the extraction of the corpus data, the filtering of the corpus data was conducted manually. Two kinds of concordance lines were excluded in our study.
First, the wrong POS-tagging in the concordance lines were removed. For example,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the substitute in (3.1) is used as the noun modifier which presents the identity of Andy Payton as a substitute player.
(3.1) Substitute Andy Payton had a chance after Stuart Ripley went off with a dead leg. (K4T-3754)
Since the SUBSTITUTE in (3.1) is not a verb, we removed it from our analysis. In addition, we also ruled out any repeated concordance lines and kept one for analysis.
Then, according to the Oxford English Dictionary, in mathematics,
SUBSTITUTE specifically means that ‘to put (a quantity) into an equation, formula, etc., in the place of an existing quantity wherever the latter occurs’. Due to its conventional meaning in mathematics, we do not include the SUBSTITUTE in mathematics in our analysis of construal and information status. However, being the verbal SUBSTITUTE, they are counted as part of the total number.
Then, after filtering the corpus data, 1104 concordance lines (excluding the
‘mathematics’ category) were left for analysis. Among them, 95.5% of instances are in written texts (1055 hits), while only 4.5% of uses belong to spoken data (49 hits).
3.3 Data Analysis
In this section, we introduce the criterion of the annotations required to answer the research questions. First, to answer the first two research questions concerning the types and distribution of sentence pattern, the criterion of recognizing a distinct sentence pattern is introduced in section 3.3.1. Then, as for the construals underlying the sentence patterns, the annotation of the profiled participants is introduced in section 3.3.2 and the trajector-landmark alignment is in section 3.3.3. Then, in section 3.3.4, the geometrical representation of the construals in action chains will be
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
illustrated. Lastly, the hypotheses and the criterion of annotating information status will be offered in section 3.3.5.
3.3.1 Recognition of Sentence Patterns
In the present study, the recognition of sentence patterns follows Langacker’s (2008) notion that language is “symbolic in nature” (p. 5). Simply put, language is a system constituted by the form-meaning pairs. This view is shared in the community of Cognitive Linguistics that “syntactic structures at various levels of schematicity are considered meaningful in their own right” (Croft & Cruse, 2004, as cited in Gries &
Stefanowitsch, 2007:2). This view denies the postulate in Generative Grammar (Chomsky, 1986) that the surface form of language is derived from the deep structure through the process of transformation. For example, the surface form of the passive construction (see 3.2a) is considered the result of transformation from deep structure of (3.2b). In other words, in Generative Grammar, the passive construction does not have its own right but a derivative surface form of the active transitive form.
(3.2) a. [OUT The drugs] was substituted [IN with another substance] [AGENT by Customs officers].
b. [AGENT Customs officers] substituted [OUT the drugs] [IN with another substance]. (K4M-734)
However, as we follow the view that language is the form-meaning pairs in nature, we recognize that the passive construction of (3.2a) is meaningful and has its own right. In other words, (3.2a) is the passive construction distinct from the active transitive construction of (3.2b). Then, in the passive construction, (3.2a) forms the more specific sentence pattern in the brackets: [NPOUT + be + Verb-pp + with NPIN +
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
by NPAGENT]. That is, the sentence pattern [NPOUT + be + Verb-pp + with NPIN + by NPAGENT] is one of the manifestations of the passive construction. In contrast, (3.2b) forms the sentence of [NPAGENT + Verb + NPOUT + with NPIN] as one of the
manifestations of the active transitive construction. In addition, the tense and aspect of SUBSTITUTE are considered negligible, and thus the past tense in (3.2c) is neither labeled in the sentence pattern nor recognized to form a distinct sentence pattern.
3.3.2 Annotation of the Profiled Participants
We briefly recapitulate Langacker’s (1991, 1999, 2008) proposal of profiling as follows. First, the frame (or base) entails the knowledge and experience of the event, such as the commercial transaction event. Then, to describe the experience of the event, language users encode different expressions by profiling different frame elements (i.e. participants) in the frame (i.e. the base) to direct the attention. By making the profiled participants prominent and leaving others unspecified in the base, a certain expression entails the conceptual content and against which a particular construal is imposed, that is, the participants that language users direct attention to.
The notion of ‘profiling’ is applied in our analysis of SUBSTITUTE, and the
‘base’ which SUBSTITUTE is profiled against is the {REPLACING} frame, in which three core FEs, ‘AGENT’, ‘IN’, and ‘OUT’(seeTable 2.2, repeated here as Table 3.1), could be profiled to encode the scene as the profiled participants. In this thesis, the terms ‘participant’ and ‘frame element’ are used interchangeably, and both refer to the required element of the conceptual base.
In this thesis, the terms ‘participant’ and ‘frame element’ are used
interchangeably, and both refer to the required element of the conceptual base.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 3.1 The definitions of the {REPLACING}frame and its core FEs Frame REPLACING
Definition An ‘AGENT’ changes the filler of a role by placing an ‘IN’ filler in the position after the ‘OUT’ filler ceases to occupy the position.
Core FEs ‘AGENT’ The conscious entity, generally a person, that performs actions resulting in the ‘IN’ entity occupying the position.
‘IN’ The ‘IN’ entity is the person or thing that the ‘AGENT’ sets to fill a role
‘OUT’ The entity that formerly occupied the position.
Then, since SUBSTITUTE is a verb which profiles a ‘relationship’ (or ‘relational profile’ in Taylor, 2003:205), which designates the relation between core FEs in the {REPLACING} frame. In other words, SUBSTITUTE profiles the relationship, specifically the interaction between the core-FEs in the {REPLACING} frame. For example, in gas can substitute petrol, SUBSTITUTE designates the relation between the first ‘thing’ (the ‘IN’, gas) and the second ‘thing’ (the ‘OUT’, petrol).
The annotation of the profiled participants conforms to the definition of the core FEs in the frame (see Table 3.1). In concordance lines, a given segment would be considered the realization of the profiled participant if the segment fulfills the
definition of the participant. The annotation of participant is labeled in the subscript of the particular segment. For example, in (3.3), the NP you is the agent performing the action of substituting, and thus ‘AGENT’ is labeled in the subscript of you.
(3.3) [AGENT You] can substitute [IN lime juice] [OUT for lemon juice] in a recipe.
Then, since the NP (lime juice) is the entity that the ‘AGENT’ sets to replace the ‘OUT’, it is labeled with the subscript of ‘IN’. As for the prepositional phrase (for lemon
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
juice), since the lemon juice is the entity being replaced and it is realized in the prepositional phrase, the for-phrase is labeled with ‘out’. The realization of the three participants makes ‘AGENT’,‘IN’, and ‘OUT’ are the three profiled participants in (3.3);
put it in another way, ‘AGENT’,‘IN’,and ‘OUT’ are profiled in (3.3).
However, some difficulties were encountered. Given that the concordance lines of SUBSTITUTE came from a wide variety of domains (e.g., sports news and
chemistry), the recognition of the profiled participant in the concordance lines may be difficult. If the interpretation of the concordance lines requires less technical
knowledge of a specific domain (e.g., sports news), we brought the concordance lines to the members of the Corpus-based research group at National Chengchi University for further examination. Then, as for those requiring some technical knowledge, we looked up the Oxford English Dictionary (OED) for reference and consulted the resources with the specialties in the specific domain. For example, (3.4) was found in the chemistry domain relating to the substitution between atoms.
(3.4) Olah, however, managed to perform Friedel-Crafts reactions on these polychlorinated fullerenes, which showed that at least 22 phenyl groups had been substituted. (ALW-360)
However, according to the OED, the entry of the verbal SUBSTITUTE in chemistry could be either “to put (an atom or group) in the place of an existing atom or group in a molecule” or “to replace (an existing atom or group in a molecule) with a different one”. In other words, the NP in the subject position (22 phenyl groups) could be either NPIN or NPOUT. Now that the entry of the OED is less helpful in distinguishing the role of the NP in question, we turned to the resources with the specialty in chemistry and consulted the online resources, such as Wikipedia, for
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
reference.12 With these resources, we could be certain that the NP (22 phenyl groups) in (3.4) is NPOUT.
The other difficulty relates to the annotation of the profiled participants realized by the prepositional phrases. This difficulty comes from the polysemous nature of the prepositional phrases. For example, in (3.5a) the for-phrase indicates the ‘reason’
rather than the ‘OUT’ participant in the {REPLACING} frame. Similarly, in (3.5b) the with-phrase indicates ‘having or including something’ rather than the ‘IN’ participant.
(3.5) a. But the jury accepted that a page of the statement had been substituted for legitimate reasons. (K23-954)
b. These are my preferences, but many other professional art materials could be substituted with similar results. (CFL-516)
Therefore, to annotate the prepositional phrases accurately, we did not rely on the forms (e.g., for-phrase) alone but examine the contexts. Some of the concordance lines were discussed in the corpus group for further examination if they were difficult to pin down.
3.3.3 Annotation of Trajector-Landmark Alignment
Different degrees of prominence are conferred on the profiled participants to reflect the particular construal of language users. As reviewed in last chapter, the primary focus, trajector (TR), is conferred on the subject of the verb, and the secondary focus, landmark (LM), is conferred on the direct object of the verb. In addition to the ‘focal participants’ as the trajector or landmark, the profiled participant
12 URL: https://zh.wikipedia.org/wiki/%E5%82%85-%E5%85%8B%E5%8F%8D%E5%BA%94
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
in the oblique (e.g. prepositional phrase) is the secondary landmark (lm) in the clause.
We label the prominence of each participant in the superscript of the corresponding segment. For example, in (3.4), the ‘AGENT’ (Customs officers) is the syntactic subject of the verb, and it receives the prominence of being the trajector in the clause. The annotation of trajector is labeled in the subscript of the NP you. Then, the direct object (the drugs) is the primary landmark of the clause, and the landmark (LM) is labeled in the subscript as well. Lastly, the oblique (with another substance) is the secondary landmark in the clause with the label ‘lm’ in the subscript of the prepositional phrase.
(3.4) [AGENT Customs officers]TR substituted [OUT the drugs]LM [IN with another substance]lm. (K4M-734)
The trajector-landmark alignment in (3.4) reflects the construal in which the
interaction between ‘AGENT’ and ‘IN’ is the focus, while the ‘OUT’ participant is made relatively peripheral by realizing it in the oblique.
3.3.4 Diagram as the Representation of Construal
As indicated by Croft (2012:6), “the semantic representations in cognitive linguistics are intended to represent the conceptual structure rather than truth conditions in the world” (p. 6). We follow the tradition of cognitive linguistics that
“semantic representations tend to be diagrammatic” (Croft, 2012:6) by drawing the diagram of the conceptualized ‘action chain’ (Langacker, 1991, 1999, 2008).
Following Langacker (1991, 1999, 2008), we draw the diagram of (3.4) in Figure 3.2 for illustration. In Figure 3.2, each circle represents a participant in the action chain. If the circle is bold, it is profiled in the conceptual base, and thus the three circles are all bold in Figure 3.2. Then, a solid arrow represents the force vector, that
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
is, the flow of energy in the action chain. In Figure 3.2, the ‘AGENT’ exerts force to the
‘IN’ participant, which in turn transfers the energy from the ‘AGENT’ and affects the
‘OUT’. Also, the dashed arrow represents the possible change of state or location in
‘OUT’ without affecting other participants. While ‘AGENT’ and ‘OUT’ are the ‘head’
and the ‘tail’ of the action chain, the ‘IN’ participant is in the central position used as an instrument which affects the ‘OUT’ by taking the place of it. We take the ‘head’ and the ‘tail’ of the action chain, and call it the ‘AGENT-OUT action chain’.
Figure 3.2 Diagram of the ‘AGENT-OUT action chain’
Therefore, in (3.4), it describes the event that the ‘AGENT’ (Customs officers) uses the
‘IN’ (another substance) as the instrument to replace the ‘OUT’ (the drugs). However, the attention is particularly directed to the portion that the ‘AGENT’ replaces the ‘OUT’.
However, the participants are not always included in the energy transmission. In (3.5), the interaction between the ‘AGENT’ and ‘IN’ is the focus, while the ‘OUT’ is realized in the oblique (for lemon juice). Despite the equated number of the profiled participants with (3.4), different construals imposed on the event give rise to different action chains. The action chain of (3.5) is diagrammed in Figure 3.3.
(3.5) [AGENT You]TR can substitute [IN lime juice]LM [OUT for lemon juice]lm in a recipe. (CEK-4699)
Agent In Out
TR lm LM
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
In Figure 3.3, the ‘AGENT’ exerts force to the ‘IN’ participants, as in Figure 3.2, but the ‘IN’ participant does not transfer the energy from the ‘AGENT’ to the ‘OUT’ (diagramed in the dashed line). Instead, the dashed line indicates the ‘OUT’ is merely relevant to the ‘IN’ participant in the sense that the replaced entity is specified. In other words, in Figure 3.3, ‘AGENT’ and ‘IN’ are the ‘head’ and ‘tail’ of the action chain, while ‘OUT’ is merely relevant to the ‘IN’ participant by specifying the replaced entity. The ‘OUT’ participant does not hold the force-dynamic relation with the ‘IN’ participant, and thus it is not part of the action chain. We take the ‘head’ and the ‘tail’, and call the action chain in Figure 3.3 the ‘AGENT-IN action chain’. It reflects the construal which focuses on the interaction between the ‘AGENT’ and the ‘IN’, while the ‘OLD’ is relatively peripheral.
Figure 3.3 Diagram of the ‘AGENT-IN action chain’
By comparing Figure 3.2 and Figure 3.3, we can see that different action chains can be conceptualized in the event of substituting. Then, different degrees of
prominence are conferred on profiled participants, and in turn represent distinct construals in each action chain. For example, the event of substituting in (3.5) can be realized in another sentence pattern representing different construal, as in (3.6).
(3.6) [IN Lime juice]TR can be substituted [OUT for lemon juice]lm in a recipe.
Agent In Out
TR LM lm
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
While (3.5) and (3.6) share the ‘AGENT-IN action chain’ in Figure 3.3, they differ in the prominence conferred on the participants. In (3.6), the primary focus is conferred on the ‘IN’ participant, the ‘OUT’ is still the secondary landmark, and the ‘AGENT’ is not profiled. The varying degree of prominence in participants gives rise to distinct construal of an action chain.
However, note that the conceptualization of the event of substituting in different action chains is also a matter of construal. The action chains reflect how language users construe the event of substituting. Specifically, in the ‘AGENT-OUT action chain’, the event is conceptualized that an ‘AGENT’ uses ‘IN’ as an instrument to replace the
‘OUT’. However, in the ‘AGENT-IN action chain’, the event is conceptualized that an
‘AGENT’ chooses an entity as an ‘IN’, which occupies the former position of ‘OUT’. No force-dynamic interaction occurs between ‘IN’ and ‘OUT’ in ‘AGENT-OUT action
chain’.
In the present study, we draw the diagram for the representation of different construals which encompass an action chain and the degree of prominence conferred on profiled participants.
3.3.5 Annotation of Information Status in Information Structure
Information structure is argued to be potential in distinguishing the role of the particular NP, especially the information status of the NP in the direct object position (e.g., You can substitute margarine in the recipe) and the NP in the subject position (e.g., Margarine can be substituted in the recipe). The potential, we argue, comes from the constraint that we can hardly regard new information as the NPOUT. Instead, in the event of substituting, new information tends to be the NPIN as the alternative.
For example, imagine the scenario of discussing the ingredients of a meal in the recipe. It is common to use (i.e. NPIN) or remove (i.e. NPOUT) the ingredient that
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
originally exists in the recipe as old information. Then, during the discussion,
someone proposes to use a new ingredient (i.e. NPIN) that does not exist in the recipe and thus new information to everyone. However, we can hardly introduce a new ingredient while at the same time removing it as an NPOUT. In fact, it would be absurd to claim the newly introduced ingredient (new information) as the NPOUT. In general, it could be more intuitive to propose something new in the discourse as an NPIN rather than an NPOUT. This constraint demonstrates the potential of information status to distinguish the role of NP in SUBSTITUTE, and two hypotheses are proposed as follows. The first hypothesis predicts that the NP being new information tends to be NPIN rather than NPOUT. Then, the second hypothesis predicts that the NP which is old information shows the neutral preference to either NPIN or NPOUT since the constraint of the ‘discourse-new’ NP as indicated above is precluded.
To attest these two hypotheses, the annotation the information status of the NP in question is needed. In section 2.3.1, we have reviewed the taxonomy of information status proposed by Prince (1992:309). However, we should exercise caution with the context where the taxonomy is used. In particular, the taxonomy of information status in Prince (1992) is mostly applied in conversations between interlocutors, that is, the spoken language. Apparently, this differs from the 95.5% written language in our
To attest these two hypotheses, the annotation the information status of the NP in question is needed. In section 2.3.1, we have reviewed the taxonomy of information status proposed by Prince (1992:309). However, we should exercise caution with the context where the taxonomy is used. In particular, the taxonomy of information status in Prince (1992) is mostly applied in conversations between interlocutors, that is, the spoken language. Apparently, this differs from the 95.5% written language in our