CHAPTER THE CONTEMPORARY DEVELOPMENT OF CONCEPTUAL METAPHOR THEORY 2.1. Conceptual Metaphor Theory: Origins and development Conceptual Metaphor Theory (henceforth CMT) is the dominant paradigm in metaphor studies. The school of cognitive metaphor analysis founded by Lakoff & Johnson (1980) claims that “metaphors play a central role in the construction of social and political reality”. According to this paradigm, Conceptual Metaphors are defined as systematic sets of correspondences or “mappings” across conceptual domains whereby a more abstract “target” domain is partly structured by a more concrete and experiential “source” domain. Thus, CMT in the Lakoff & Johnson tradition views linguistic metaphors and similes as reflecting mappings across domains of knowledge that undergird the language user’s understanding of the world in which they live. Lakoff & Johnson go as far as to claim a strong version of CMT that sees metaphorical thought in the form of cross-domain mappings as “primary” whilst metaphorical language is “secondary”. According to the Conceptual Metaphor paradigm, examples of general metaphors include pathmovement-journey metaphors (“GOOD PROGRESS IS FORWARD MOVEMENT”) and the strict father vs. nurturing parent model (“MORAL AUTHORITY IS PARENTAL AUTHORITY”, “A COMMUNITY IS FAMILY” etc.). The basic idea is that these Conceptual Metaphors have entailments through which they highlight and coherently organise those aspects of our experience through an 21 experiential or embodied dimension. Thus, in their view, metaphors may “create social realities for us and become a guide for future action” (Lakoff & Johnson, 1980). A good example to illustrate this point would be the “ARGUMENT IS WAR” metaphor where many aspects of arguments are framed in terms of war e.g. “shooting down arguments”, “launching counter-attacks”, winning strategies that are “right on target” etc. However, a key criticism of CMT is that it is founded on a relatively small set of samples that were either artificially constructed or arbitrarily collected. Thus, one major proposition to rectify this would be through the adoption of a corpus methodology. When used in tandem with a systematic and theoretically-sound framework, this has the potential to put CMT on a sounder empirical and academic footing that enables the extrapolation of conceptual metaphors from authentic linguistic data. This thesis aims to extend existing corpus methodology in the identification and analysis of conceptual metaphors in large datasets through the use of semantic annotations in WMatrix software and word sketches in Sketch Engine as an empirical, inductive springboard. But before we move onto the introduction of the new proposed framework, it is important to acknowledge the key shortcomings plaguing Conceptual Metaphor Theory since its inception. The original assumption by Lakoff & Johnson (1980) that metaphor is a cross-domain mapping in the conceptual system is being challenged by a variety of theorists – with the key points of contention coming from Class Inclusion Theory (Glucksberg & Keysar, 1990; Kintsch, 2000; Jones & Estes, 2005) and Relevance Theory (Wilson & Sperber, 2002, Moreno, 2005). In response, Gentner & Bowdle’s (2008) Career of Metaphor Theory attempts to integrate the main concerns 22 raised by arguing that the mental processing behind individual metaphors “evolve” over the course of their lives from comparison (cross-domain mapping) in the early stages, to categorisation when they become conventionalised (p.123). So this means that the standard source and target domain mapping proposed by Lakoff & Johnson (1980) is increasingly viewed as over-simplistic as a variety of cognitive mechanisms tend to work in tandem in order to ensure cognitive efficiency for different metaphors, both conventional and novel. The main claim is that the Career of Metaphor Theory offers a sophisticated description of the relation between the cognitive and linguistic forms of metaphor. On the other hand, Fauconnier & Turner’s (1997, 2002, 2008) Conceptual Blending Theory further claims that metaphorical thought is best represented as a higher level of organisation rather than a simple cross-domain mapping. This means that metaphorical thought comprises a network of connected mental spaces that results in emergent structure created by the blended space and the entire network. Conceptual Blending (also known as the “network model”) states that these blended spaces are often structured by more than one conceptual domain or input spaces and these result in a fourth generic space that makes the sharing of abstract structures possible (Kovecses, 2002, p.227-232). This highlights the importance of the socio-cultural context as a key-determining factor in the cognitive processing of metaphors, apart from the usual linguistic and conceptual considerations. These recent developments in metaphor theory will be taken into account in the proposed formulation of the new discourse-analytical framework that attempts to capture metaphor in language, thought and communicative context. This 23 simply means that the representation of the conceptual metaphors mined from the corpus will include a socio-cultural dimension in addition to the usual linguistic and conceptual considerations in its interpretation and presentation. 2.2. Previous Studies: Main Trends This segment aims to provide a chronological overview of the contemporary development of conceptual metaphor theory from an amalgamation of selected metaphor theorists. While this overview is by no means a comprehensive listing of all the significant works by metaphor theorists, it forms the broad source of inspiration as well as the theoretical springboard for subsequent development of the Integrated-approach to the Identification of Conceptual Metaphors (IICM) proposed in this thesis. 2.2.1. An Inductive Approach to the study of Metaphor (Partington, 1998) Partington’s (1998) key contribution to CMT lies in three main areas. Firstly, he advocates a distinction between Systematic Metaphors or Source Metaphors and Genre-Typical Metaphors. Partington notes that Lakoff & Johnson’s metaphorical concepts e.g. “THEORIES ARE BUILDINGS”, “IDEAS ARE PLANTS”, “TIME IS A VALUABLE COMMODITY”, fall under the category of a “semantic mega-metaphor”. His main criticism is that Lakoff & Johnson pay no attention to the notion of genre or text-type. Thus, Partington proposes that data from corpora should be accessed according to a number of different 24 mega-genres or text-types e.g. according to different sections of the newspaper and texts of miscellaneous disciplines in the academic genre in order to provide an element of representativeness and generalisability to the findings and interpretations. He feels this distinction ought to be made as metaphor is “sometimes serious, at others self-conscious and tongue-in-cheek” according to the communicative context. Thus, Partington highlights the significance of consistency in the topic and text-type for metaphorical analysis. He proposes that the analysis has to be taken in conjunction with social, historical and contextual factors in the identification and formulation of conceptual metaphors. His second key contribution would be his support for the use of authentic data to support theoretical arguments on Conceptual Metaphors. He advocates the amalgamation of a corpus-linguistic approach together with CMT. For example, Partington feels that in terms of methodology, it is not sufficient to refer to the relevant sections of a newspaper in isolation. Hence, the significance of a metaphor finding must be gauged empirically against a reference corpus. This is because it is important to make use of a reference corpus to make a point on the significance of the findings in the targetted text types through a cross-corpora comparison of the nature and frequency of metaphors. This approach advocates an empirical foundation for the formulation of conceptual metaphors that is based on authentic language data that is not artificially elicited. Thirdly, Partington supports an inductive approach to data analysis. He feels that it is paramount to let the data itself suggest the range and types of metaphors that might be found in a corpus rather than guessing a priori what 25 these might be. Hence, he sees the first step to metaphorical analysis as necessarily manually parsing the texts through ocular scanning. It is clear that Partington advocates a brand of Conceptual Metaphor Analysis that is genrespecific and supports qualitative analysis with quantitative data on metaphor frequency. This provides a more robust account of the discourse role of metaphor. Partington’s theoretical stipulations are encapsulated within this thesis in the choice of genre (online newspapers), the delimitation of the specific topic under study (the BP Oil Spill in 2010 ranging from 20 April – 20 July), the conscious use of the informative writing section of the BNC sampler within WMatrix as a reference corpus as well as the use of USAS semantic categories to mine the relevant metaphors systemically and inductively rather than utilising the traditional method of using intuitively pre-determined search strings. 2.2.2. Methodological and Theoretical Issues in Researching Metaphor (Cameron & Low, 1999) This book comprises a collection of landmark articles by prominent contemporary metaphor theorists with the main aim of developing an approach towards conceptual metaphor theory that goes beyond mere theoretical semantics and involves a “delicate integration of word-sense, syntactic form, pragmatic context, speaker-listener relationship and goals” (p. xiii). Cameron (1999) focuses on HOW to operationalise the concept of metaphor for research of an applied linguistic nature. Hence, she advocates a 26 view of language where the social and cognitive aspects are integral to formulation of theory and the analysis of data. Such an approach foregrounds the salience of temporal and contextual factors in the process of metaphor formulation. Cameron also differentiates between two different levels of metaphorical analysis: 1. The Theory Level of Analysis focuses on metaphor identification, the categorisation of metaphor types as well as the goals and logic of production and interpretation (p.7). 2. The Processing Level of Analysis studies the interaction between the individuals and their socio-cultural environments in processing metaphorical language in specific discourses as well as how metaphor processing can alter conceptual structures and imbue new meanings to lexical items (p.6). Hence, Cameron asserts that all conceptual metaphor analysis must take place in a framework that specifically accommodates different metaphor types and considers the language in a goal-directed discourse context. This discourse context should be divided into a local, discourse and global context of systematicity. Thus, Cameron emphasises the importance of amalgamating the socio-cultural contexts as well as the relevant discourse goals in the formulation of conceptual metaphors that is culturally relevant and comprehensive. Gibbs (1999) proposes several guidelines in researching and applying metaphor. One of his key ideas is the need for researchers to acknowledge that 27 their accounts of metaphor are only limited to particular TYPES of metaphor. Here are Gibbs’ proposed categories for metaphor classification (p.30-36): 1. A IS B/ A IS LIKE B: Refers to the typical cross-domain metaphors and similes in the traditional form of conceptual metaphor. 2. XYZ metaphors (p.31): Refers to proverbial expressions e.g. the love of money is the root of all evil; religion is the opiate of the masses; wit is the salt of conversation. 3. Image metaphors (p.32-34): Used in contexts where mental images from one source of knowledge are mapped onto another (rather than inter-domain concept mapping). e.g. my wife whose hair is brush fire, whose thoughts are summer lightning, American civilisation is like an orchestra (the symphony of civilisation) etc. 4. Single metaphors (p.34-35): These are metaphors that reflect a particular entailment of the conceptual metaphor - LOVE IS A JOURNEY. e.g. we are at crossroads, our marriage is on the rocks, the relationship is a dead-end street, we can’t turn back now etc. 5. Metonymy (p.36-37): Metaphor involves mapping across two different conceptual domains whilst Metonymy involves only transfer within one conceptual domain. Finally, Gibbs also emphasises the importance of recognising the experiential dimension of metaphors for the twin communicative aims of vividness and compactness. Hence, the types of metaphorical manifestations as well as the 28 contextual and cultural dimensions of the issue under analysis will be fully analysed and examined in this thesis. Low (1999) identifies the vital need for researchers to ensure that the data, theoretical framework and the associated methodologies are representative, suitable and consistent in relation to the topic under study. He emphasises that the researchers should not oversimplify and generalise the inbuilt complexity of conceptual metaphors. I will highlight two key issues in validating metaphor research that are applicable to the concerns in this thesis. 1. Low identifies the inherent subjectivity in current methods of metaphor identification by individual researchers. This subjectivity is due to discrepancies in conceptual metaphor identification criteria as well as the possibility of over-interpretation due to researcher bias and the phenomenon of “metaphor hypersensitivity” (p.53). Low claims that “post-hoc questioning” often leads to artificially increased information density. This is the reason why a standardised conceptual metaphor identification procedure that is theoretically-sound needs to be formulated and continually refined. 2. Secondly, it is vital to use a representative corpus within a delineated context for the study of metaphors in order for the findings to be validated accordingly. Hence, it is clear that the predisposition and the interactions within the target group in the discourse influences the interpretation of metaphors by the researcher. This is a key consideration in the selection of the respective 29 broadsheets from America and Britain respectively, where the target readers will perceive themselves either as victims or as wronged perpetrators depending on their respective end of the spectrum. This will be coupled with the proposed development of a semi-automated metaphor identification procedure from a large representative corpus in order to ensure increased metaphorical validation by reducing the element of researcher bias. 2.2.3. Metaphorical Entailments and Political Legitimation – (Chilton, 1996; 2004) Chilton (1996) examines the use of metaphor in the political domain and claims that political leaders are viewed as persuasive when their metaphors succeed in legitimising policies. Thus, political leaders use metaphors that will either present their own policies in a positive light or will disparage those of their opponents through the media. This makes metaphor an ideological tool for shaping the political or media landscape. Chilton notes that a news article evokes its most “compelling cognitions” through the metaphoric views it takes for granted rather than those it explicitly asserts (p.36). Chilton adopts a diachronic, top-down deductive approach rather than a bottom-up inductive approach in examining the three concepts of security, state and sovereignty in Cold War discourse. He also examines how these abstract concepts are constructed and mediated through the passage of time and the changing historical landscape. Chilton has two overarching aims in this research endeavour: 30 It is important to note that Steen’s groundbreaking method, while timeconsuming, provides a systematic and well-documented procedure for the identification of conceptual metaphor from linguistic manifestations. This is key in order to prevent linguists from seeing “concrete manifestations of conceptual metaphors everywhere” (Steen, 2009, p.215). This highlights the importance of an inductive approach to the study of conceptual metaphors because if a conceptual metaphor is pre-determined, the analyst will inevitably find the kind of evidence he/she is looking for, overlooking potentially salient alternative mappings, compromising on the potential metaphorical richness of the data. This simply means that it is vital for the conceptual metaphors to be formulated on the basis of concrete manifestations of linguistic evidence in the discourse. However, while Steen’s proposed approach is systematic and more comprehensive, Steps and remain inaccessible due to the need for a formal grasp of the concept of “propositionalisation” and a concrete understanding of the formulaic representations. Steps and remain potentially limited by the factor of subjective intuition – constrained by the social, historical and cultural contexts of the discourse and the analyst’s viewpoint. Hence, a whole new 4step process broadly inspired by (but not based on) Steen’s groundbreaking contribution will be counter-proposed for the IICM in Chapter 3. This new 4step process can be seen as a more “accessible” framework that is empirically based on WMatrix and Sketch Engine, two separate integrated web-based concordancing programmes, as springboards for analysis. Furthermore, the incorporation of the word sketch function in Sketch Engine further systematises the conceptual metaphor formulation procedure by including the 53 consideration of a word’s grammatical classification in tandem with its collocational behaviour. 2.2.10. The Semi-Automated Mining of Metaphors from Large Representative Corpora (Hardie, Koller, Rayson & Semino, 2007) Apart from the research conducted by Hardie, Koller, Rayson and Semino (2007), Ooi et al. (2007) have adopted the integrated corpus linguistic tool, WMatrix, in a systematic analytical approach that examines and fosters a greater awareness of the “creative acts” in Internet discourse among Singaporean teens and undergraduates. Their key argument is that a close study of such creative linguistic patterns can lead to a deeper understanding of the various cultural identities at play. This is in line with the objectives of this thesis where the influence of situational, sociocultural and political context on the choice of metaphors employed in the respective national broadsheets is examined. In a separate paper, Ooi (2008) embarked on a cutting-edge project involving the examination of the lexis in online electronic gaming, once again using WMatrix. Ooi describes WMatrix as a “leading corpus linguistic software” and uses it as a platform to analyse a 1-million word “G-corpus” dataset. This dataset comprises sources of gaming language such as manufacturer’s product descriptions, gaming reviews, weblogs and discussion forums. Ooi’s project focuses on the use of WMatrix as a springboard to adapt and refine Sinclair’s (2004) posited lexical model consisting of semantic prosody, semantic preference, colligation, collocation and the core lexical item. This research serves as an excellent precursor to this present study as it 54 demonstrates the utility of WMatrix in the analysis of authentic linguistic phenomena - extending from a lexical platform to a more evaluative semantic prosody that links to the specific discourse contexts. The research conducted by Hardie et al. (2007) describes the exploitation of WMatrix for the “identification of metaphorical patterns in large datasets”. The proposed methodology by Hardie et al. (2007) involves selecting and analysing possible candidate semantic fields proposed by USAS (UCREL Semantic Annotation System) to assist in finding possible “source” (e.g. JOURNEY, WAR) and “target” (e.g. LIFE, ARGUMENT) domains. The USAS tagger provides for the automatic semantic annotation of text. This allows the researcher to approach the text from a bottom-up, inductive perspective as the annotations are accomplished by the WMatrix programme. According to Rayson (2008), USAS shows “semantic fields which group together word senses that are related by virtue of their being connected at some level of generality with the same mental concept”. In other words, it consists of a semantic tag-set embodying a general language ontology loosely based on Tom McArthur's Longman Lexicon of Contemporary English (Archer, Wilson & Rayson, 2002). The tag-set distinguishes 21 major semantic fields and 232 subdivisions (See Figure 2.6). Once the semantic tags have been assigned, the researcher can then go on to consider the “potential metaphoricity” of the selected lexical items by examining the lexical concordances available under the USAS interface within WMatrix based. This is accomplished via the deliberate identification of semantic fields that not correspond with the central topic of the data under analysis. These “discordant” semantic fields will potentially form the candidate source domains from which the analyst can 55 identify a fuller range of metaphorical expressions than through the traditional use of pre-determined search strings. It is important to note that these candidate semantic fields (as utilised under Hardie et al.’s methodological paradigm) are not necessarily empirically salient on the list of overused items in a particular corpus. This method thus enables the researcher to inductively and systematically identify “open-ended sets of metaphorical expressions” which are not limited to “pre-determined search strings”. As the USAS semantic domains exhibit a rough correlation with source and target conceptual domains, this framework will serve as a good platform for the analysis of the metaphorical framing of the conflict over the Gulf of Mexico Oil Spill through an examination of potential source and target domains for further metaphorical analysis. Hardie et al. (2007) assert that utilising a semantic tagger for metaphor analysis “yields two to three times more results than manual analysis” (p.11). While the research team acknowledges that the proposed semantic annotation method still requires “extensive manual reworking on the part of the researcher”, they claim that this process has the added advantage of generating more comprehensive results in an inductive manner that does not overlook potential metaphorical alternatives. This process is aided by the use of an additional “broad-sweep” function, which provides additional information on each lexical item according to its range of possible semantic domains. The “broad-sweep” function ensures that the search includes all lexical items with a relevant tag in their list of tags and thus, provides a platform for a thorough metaphorical analysis that is inductive and comprehensive. 56 Figure 2.6 Screenshot of the 21 USAS Semantic Domains However, it is vital to note that there are two main limitations to the use of WMatrix for conceptual metaphor analysis: 1. WMatrix merely generates concordances and semantic domains on a lexical level. This means that the lexical and semantic taggers are based on a pre-existing lexicon and lack the requisite accuracy and sophistication required in classifying newer or creative lexical items that may be contextually induced. Such ‘unclassified’ lexical items are sent to the Z99 bin within WMatrix to be mined by the researcher for further analysis as required. 2. While it generates the relevant wordlists and concordances, the metaphoricity of the individual lexical items as well as the formulation of the associated conceptual metaphors can only be determined by the individual researcher. This requires the use of a systematic and 57 sophisticated framework that ensures accuracy, systematicity and interrater reliability. Hence, this thesis aims to extend the existing use of integrated online concordancing programmes like WMatrix and Sketch Engine for purposes of metaphor identification and analysis. The aim is to come up with an inductive, bottom-up approach to the formulation of conceptual metaphors through the development of an entirely unique framework that uses both these platforms as springboards for further analysis. This results in an added dimension of systematicity as the WMatrix is designed for the analysis of a larger collection of representative texts and can offer empirical suggestions for the lexical items for analysis based on frequency of overuse and its associated semantic tags. Such a methodology that incorporates the “computer assisted analysis of metaphorical patterns in discourse” has the potential to systematise the point of entry for metaphorical analysis in a large swath of texts. 2.2.11. A Unified Approach (Krennmayr, 2011) Krennmayr’s significant contribution to metaphor analysis lies in her refinement of the MIP into the MIPVU (Metaphor Identification Procedure Vrjie University). This is done through the systematic incorporation of dictionaries as a point of reference in determining the metaphoricity of all the lexical items under analysis in a range of contexts. She goes a significant step beyond the mere identification of linguistic metaphors by incorporating Steen’s (2009) five-step approach to include the element of conceptual 58 metaphor formulation. Hence, her incorporation and refinement of the existing landmark theories in the field of conceptual metaphor analysis have resulted in a unified approach that injects an element of systematicity and empirical rigour to the identification of conceptual metaphors. Krennmayr also explores the use of the USAS domains within the WMatrix as a springboard for semi-automated metaphorical analysis. In her view, the USAS semantic domains bridge the gap between lexis and semantics and by extension, have the potential to “bridge the linguistic and conceptual domains” (p.236). She proposes two methods for the use of the USAS semantic domains and differentiates them into a “top-down approach” and a “bottom-up approach”. This thesis aims to extend Krennmayr’s “top-down” approach on a much larger scale. She labels the use of USAS domains as a springboard a “top-down approach” due to the pre-allocation of semantic fields by WMatrix even before any analysis takes place. However, this approach can be alternatively defined as a “bottom-up approach” as the USAS domains merely provide an indication of the semantic categories that stand out from the reference text and thus, does not involve the researcher in the initial categorisation and selection of the lexical units for metaphorical analysis. The conceptual metaphors will only be formulated from a detailed analysis of the relevant lexical units and their respective concordances. In other words, the USAS domains only serve as a springboard for further metaphorical analysis. While Krennmayr has done a preliminary exploration of WMatrix’s potential for metaphor analysis, this thesis explores another methodological route via the formulation of the Integrated-approach to the Identification of Conceptual Metaphors (IICM), a methodological framework that is broadly 59 inspired by but is distinct from current frameworks. The metaphorical conceptualisation of the fallout between America and Britain in the BP oil spill tragedy will be the conduit for the demonstration of this methodology that essentially proposes a semi-automated procedure for conceptual metaphor analysis. Krennmayr criticises the notion that the USAS domain classifications is a “quick and fast method of finding source domains and their corresponding metaphorical expressions” (p.201). This is due to the potentially timeconsuming process of analysing the respective semantic domains in a comprehensive manner. Her preliminary research has shown that the fields that not appear to have the potential for metaphoricity are often rich sources of conceptual metaphors. Thus, while Krennmayr’s assessment holds true to a certain extent, it is undeniable that WMatrix, together with the use of the broad-sweep function remains an excellent springboard and a potentially comprehensive tool for metaphor analysis in large collections of representative text if it is utilised with the right framework. A preliminary analysis of the data using the IICM in Chapter shows that it supports Hardie et al.’s (2007) claim that it unearths more conceptual metaphors than through traditional, intuitive manual analysis. 2.2.12 The Case for A Unified Contemporary Discourse-Analytical Methodology (Oster, 2010; Steen, 2011) Apart from the methodological concerns in the identification and formulation of conceptual metaphors as outlined in the preceding sections 60 (Krennmayr, 2011; Steen, 2009; Pragglejaz Group, 2007; Hardie et al., 2007; Stefanowitsch, 2006), several scholars have raised relevant issues and have proposed alternative approaches to the study of conceptual metaphor theory in recent times. Oster (2010) conducted a very relevant precursor to this thesis in his study of the linguistic expression of emotions. He explores the range of possibilities and difficulties in the use of corpus methodology for the semantic and pragmatic analysis of large corpora. Specifically, he aims to address the issue of applying corpus linguistic concepts like collocations, semantic preferences and prosodies in order to “establish the conceptual metaphors and metonymies that structure our understanding of emotions (p.728)”. Oster advocates a corpus-based approach to the analysis of linguistic metaphors in natural discourse that is systematic and inductive (p.728). Oster raised three salient methodological concerns in his study. 1. Firstly, he raises the issue of establishing a “coherent methodological criteria” for the systematic, repeatable and empirical pre-selection of concordances for analysis in a large corpus (p.730). 2. As a “language-use oriented researcher”, he emphasises the need to marry both the quantitative and the qualitative aspects to the corpusbased analysis of “natural discourse” or “found-data” as this approach has the propensity to reveal interesting semantic and pragmatic aspects of the formulated conceptual metaphors. 3. In using the lexical unit as the starting point of the analysis, the key limitation would lie in the fact that purely figurative or metaphorical 61 manifestations of the concept that occurs away from the vicinity of the node word will be overlooked (p.731). In other words, Oster anticipates the methodological difficulty of finding an “adequate balance between semi-automatic and manual analysis” (p.755). This is simply because while a manual analysis is necessary, it is important to harness the computational power of corpus tools for efficient, manageable and empirical analysis – especially in the case of large corpora like the ones in this study. This highlights the need for a systematic and encompassing framework that marries the strengths of computational power with the depth of insight provided by the individual analyst. Furthermore, Steen (2008; 2011) suggests the use of a ThreeDimensional Model for the discourse-analytical study of conceptual metaphors. This discourse analytical approach simply advocates the amalgamation of the linguistic or “semiotic approach”; which incorporates the textual aspect of the analysis, with the behavioural or “social approach”. The latter in turn includes “patterns of communicative interaction” (Steen, 2011, p.46). This means that a comparative approach between the American and British broadsheets in the contextual-interpretation and discourse analysis of the metaphorical framing of the Gulf of Mexico is needed. A pictorial representation of his three-dimensional model can be seen in Figure 2.7. 62 Figure 2.7 A Pictorial Representation of Steen’s (2011) 3-Dimensional Model of Metaphors (p.19) Thus, the proposed IICM framework in this thesis aims to address Oster’s (2010) and Steen’s (2011) concerns in the following manner. 1. The systematic, repeatable and empirical pre-selection of node words for concordance analysis in a large corpus would be empirically suggested by the USAS semantic domains and the associated frequency measures. All the lexical items in the corpus would be pre-annotated by WMatrix, thus further reducing the factor of analyst intuition and pre-selection. Subsequently, the node words and concepts selected for analysis would be based on the empirical frequency of overuse as embodied by the corresponding log-likelihood values. 2. The systematic criterion-based collection of news articles from flagship broadsheets with clear socio-cultural delineations and political affiliations ensures that the data collected is natural and representative of the target population (Bednarek, 2006). Hence, we are able to enrich the description of the formulated 63 conceptual metaphors with information from a series of contextual dimensions such as intercultural exchange and mutual influences. 3. The use of the USAS semantic domains in tandem with the “broad-sweep” function as a springboard for metaphorical analysis (rather than through the sole measure of lexical frequency) minimises the chances of overlooking the purely figurative or metaphorical manifestations of the target concept that occurs away from the vicinity of the node word – especially if an exhaustive analysis of all the semantic domains is adopted. Hence, a thorough analysis of the metaphorical framing of the Gulf of Mexico Oil Spill has to be based on a representative corpus of media articles that is systematically collected in a uniform discourse event in accordance to the socio-political and geographical contexts. This can be seen in the clear separation of the British and American broadsheets with a clear delineation of the relevant political stances, both conservative and liberal. 2.3. Gaps in the Current Practice of Conceptual Metaphor Analysis The strengths of the current methods of metaphorical analysis, specifically the methods proposed by Steen (2008, 2011), Krennmayr (2011), Oster (2010), the Pragglejaz Group (2007), Charteris-Black (2004, 2005) & Musolff (2004a, 2004b), lie in the way they can help analysts identify the problematic 64 cases pertaining to linguistic metaphor and deal with them consistently and systematically. However, there are four broad areas that require further refinement in existing metaphorical frameworks. Firstly, while the intensive manual analysis of representative data samples for metaphor analysis is an entirely adequate/legitimate approach that has advanced the field of metaphor significantly, the starting point for most existing metaphorical analysis procedures remain constrained by the typical use of pre-selected search strings as the launch pad for the entire study. Since intuitions are not necessarily explicit and systematic, this can result in the procedure failing to exhaust all forms of metaphoricity in large swaths of text and thus, overlooking alternative metaphorical expressions. There is also the issue that metaphorical concepts can be found even in the absence of metaphorical language e.g. allegory (Pragglejaz Group, 2007, p.24). This simply means that socio-historical, political and other situational contextual factors as well as the co-text have to be taken into account for a thorough metaphorical analysis. Secondly, another key issue that remains contentious in most of the current literature is the subjectivity in the formation of conceptual metaphors and conceptual keys from linguistic evidence. Krennmayr (2011, p.213) has succinctly pointed out the key area of concern in the formulation of conceptual metaphors: How metaphor theorists account for the cross-classification of metaphor? In other words, how theorists decide if the source domain is one of WAR or SPORTS or PHYSICAL VIOLENCE? In their seminal work on Conceptual Metaphors, Lakoff & Johnson (1980) often postulated single conceptual mappings for their observations but not address the issue that 65 the same evidence can often be interpreted in multiple ways i.e. a single linguistic metaphor can potentially be interpreted into a range of underlying conceptual metaphors and cross-domain mappings depending on contextual, cultural or historical factors. Hence, current practices lack a widely-accepted methodology in the identification of conceptual metaphor that is comprehensive, empirical and repeatable. This thesis aims to examine the potential of using a Semantic Annotation Tool in tandem with a systematic framework in exploring new methodological possibilities in the identification, formulation and analysis of metaphors. Thirdly, there is a need for discourse-analytical studies to base their findings on carefully constructed corpora with specifically delineated topics and contexts in order to ensure validation and generalisability. This proposed study has clearly identifiable parameters that can be taken into account to explain why particular metaphors are used e.g. a clearly delineated topic/scope, a consistent socio-political context, a large sample of representative texts. While it is acknowledged that most metaphorical phenomena lack clear boundaries and often overlap, I would like to emphasise once again that this thesis focuses more on the cognitive function of metaphors rather than on the specific linguistic manifestations of metaphor and its accompanying cognitive dimensions. Hence, while there is significant existing research in the detailed study of specific metaphorical manifestations and their delineations, this study focuses on the conceptual aspects and foregrounds the function of conceptual metaphors as vital cognitive tools in organising our understanding of the world. Thus, this thesis essentially focuses on examining the conceptual metaphors that are discoursally systematic since they can be 66 seen as “the reflection of shared beliefs and assumptions of the members of particular social groups” (Semino, 2008, p.34). Fourthly, current studies often not provide a systematic account for the various metaphorical complexes used in the discourse under study. Current models seem to hint at a sprawling complex of metaphors that are loosely bound together by context and existing ontological frameworks. Current scholarly discourse in addressing this issue include Musolff’s (2004a) hierarchical organisation of “scenarios” and “conceptual metaphors” and Charteris-Black’s (2004) three-tiered organisation of “conceptual keys”, “conceptual metaphors” and “linguistic metaphors”. Taken in tandem, these four factors can compromise the authenticity of the claims and the extent to which selected target discourse elements are seen as systematically metaphorical at a local, discourse or global level (Cameron, 1999, p.16). 2.4 Summary Hence, this thesis aims to augment the current methodological approaches for analysing conceptual metaphor in a discourse analytic approach. This is done by utilising WMatrix and Sketch Engine (SkE), two separate integrated web-based concordancing tools, as platforms for metaphorical analysis in large swaths of texts. This thesis aims to come up with an inductive, bottom-up approach to the formulation of conceptual metaphors through the development of a novel and systematic framework that uses the USAS domains within WMatrix as a springboard for further analysis. This results in an added dimension of 67 empirical reliability as WMatrix is designed for the analysis of a larger collection of representative texts and will offer empirical suggestions for the lexical items for analysis based on frequency and its associated semantic tags. WMatrix also features sections of the BNC as in-built reference-corpora. This enables the calculation of the frequency of overuse through the use of Log Likelihood values (henceforth, LL values) that can be set at 6.63 corresponding to 99% significance. These LL values are then used as an empirical springboard for the selection of USAS domains and the relevant lexical items for further analysis. Furthermore, the grammatical classifications for the selected node word under analysis (and the resultant concordances generated) would result in increasingly accurate metaphorical analysis. This is because the Sketch Engine enables the analyst to go beyond examining the “arbitrary window of text around the headword” by further classifying the grammatical category of the node under study. This enables the researcher to conduct the metaphorical analysis with the grammatical relation (and the relevant collocates) of the word in mind. Thus, this provides an additional corpus interrogation strategy that further refines the study of the target discourse (Kilgarriff et al., 2004, p.3-4). Such a methodology that incorporates the “computer assisted analysis of metaphorical patterns in discourse” has the potential to provide an alternative framework to triangulate the findings of manual intensive analysis of representative data samples. This serves to marry top-down researcher intuition with a bottom-up empirical approach for the robust analysis of a large collection of representative texts. 68 [...]... formulated metaphors by highlighting what is suppressed Thus, the delineation of the topic, timeframe and the political inclinations of the broadsheet under analysis will be carefully examined and taken into account in this thesis 42 2 .2. 6 The Corpus Approach to the Analysis of Innovative Metaphors (Deignan, 20 05) Deignan (20 05) aims to find out how far a theory of metaphor and thought can account for the patterns... the advent of integrated online concordancing programmes like WMatrix (Rayson, 20 05) and Sketch Engine (Kilgarriff et al., 20 04) have made the systematic and exhaustive approach to the analysis of the empirically-salient semantic domains a realistic possibility This is simply due to the exhaustive lists of semantic domains and the associated source and target domain lexical items that are annotated and. .. tool, WMatrix, in a systematic analytical approach that examines and fosters a greater awareness of the “creative acts” in Internet discourse among Singaporean teens and undergraduates Their key argument is that a close study of such creative linguistic patterns can lead to a deeper understanding of the various cultural identities at play This is in line with the objectives of this thesis where the influence... of situational, sociocultural and political context on the choice of metaphors employed in the respective national broadsheets is examined In a separate paper, Ooi (20 08) embarked on a cutting-edge project involving the examination of the lexis in online electronic gaming, once again using WMatrix Ooi describes WMatrix as a “leading corpus linguistic software” and uses it as a platform to analyse a. .. is a proponent of Critical Metaphor Analysis (henceforth CMA) The CMA is an interdisciplinary discourse- analytic 36 approach to the study of conceptual metaphors It proposes an approach that integrates linguistic analysis with contextual understanding and social insight in explaining the use of metaphors Charteris-Black’s belief is that metaphors have three distinct yet integrated functions - a “semantic... allies; America and Britain, in online newspaper discourse will be examined in order to account for the foregrounding of certain core issues within the Gulf of Mexico Oil Spill 2. 2.4 The Discourse- Evolution of Metaphors (Musolff, 20 0 4a; 20 04b) Musolff examines the socio-cognitive dimension of metaphor by examining the phenomenon of “metaphor evolution” in a corpus of British and German debates about the. .. 3 The analysis of the findings can then be used to extrapolate the dominant conceptual metaphors in each dataset and draw conclusions 37 on the persuasive, emotive and ideological functions of metaphor in different societal and cultural domains It is important to note that in Charteris-Black’s framework, metaphors are organised hierarchically in descending order of scope in Figure 2. 2 as “conceptual... a study comprises a thorough analysis of the accompanying pragmatic factors such as the socio-cultural and communicative context as well as its persuasive functions in the associated political contexts Since metaphors are highly influential in the way they covertly shape our beliefs, attitudes and actions across social and political domains, a critical stance is needed for readers to evaluate the intentions... domains exhibit a rough correlation with source and target conceptual domains, this framework will serve as a good platform for the analysis of the metaphorical framing of the conflict over the Gulf of Mexico Oil Spill through an examination of potential source and target domains for further metaphorical analysis Hardie et al (20 07) assert that utilising a semantic tagger for metaphor analysis “yields... metaphor analysis can be accomplished through an amalgamation of a topdown and a bottom-up approach The top-down approach is done through the manual parsing of text samples within the corpus and the intuitive selection of lexical items from the concordancing programme Deignan sees that “trained researcher intuition” is indispensable for the reliable identification of conceptual metaphors from a corpus . 2. These Metaphor Keys are then concordanced in the target corpus. 3. The analysis of the findings can then be used to extrapolate the dominant conceptual metaphors in each dataset and draw. advocates a 27 view of language where the social and cognitive aspects are integral to formulation of theory and the analysis of data. Such an approach foregrounds the salience of temporal and. allies; America and Britain, in online newspaper discourse will be examined in order to account for the foregrounding of certain core issues within the Gulf of Mexico Oil Spill. 2. 2.4. The