t.c. sÜleyman demİrel Ünİversİtesİ bati dİllerİ ve ...tez.sdu.edu.tr/tezler/ts02710.pdf ·...

117
T.C. SÜLEYMAN DEMİREL ÜNİVERSİTESİ SOSYAL BİLİMLER ENSTİTÜSÜ BATI DİLLERİ VE EDEBİYATI ANABİLİM DALI METİN TUTARLILIĞININ ÇIKARIMSAL BİR MODELİ Abdul Jabbar Fadhıl Jameel AL-JAMEEL 1630224013 YÜKSEK LİSANS TEZİ DANIŞMAN Prof. Dr. Hüseynağa RZAYEV ISPARTA-2018

Upload: others

Post on 20-Oct-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

  • T.C.

    SÜLEYMAN DEMİREL ÜNİVERSİTESİ

    SOSYAL BİLİMLER ENSTİTÜSÜ

    BATI DİLLERİ VE EDEBİYATI ANABİLİM DALI

    METİN TUTARLILIĞININ ÇIKARIMSAL BİR MODELİ

    Abdul Jabbar Fadhıl Jameel AL-JAMEEL

    1630224013

    YÜKSEK LİSANS TEZİ

    DANIŞMAN

    Prof. Dr. Hüseynağa RZAYEV

    ISPARTA-2018

  • T.R.

    SULEYMAN DEMIREL UNIVERSITY

    INSTITUTE OF SOCIAL SCIENCES

    DEPARTMENT OF WESTERN LANGUAGE AND LITERATURE

    AN INFERENTIAL MODEL OF TEXT COHERENCE

    Abdul Jabbar Fadhıl Jameel AL-JAMEEL

    1630224013

    MASTER’S THESIS

    ADVISOR

    Prof. Dr. Hüseynağa RZAYEV

    ISPARTA 2018

  • iii

    ACKNOWLEDGEMENT

    I would like to express my gratitude to my advisor Prof. Dr. Hüseynağa

    RZAYEV for his invaluable guidance and useful comments throughout this study.

    I would also like to thank all my graduate course instructor; Prof. Dr. Ömer

    ŞEKERCİ, Ph. D. Carrier Önder ÇAKIRTAŞ, also amember of the advisory committe.

    My thanks and appreciation İlker ÖZÇELİK.

    Abdul Jabbar Fadhil Jameel AL-JAMEEL

    2018

  • iv

    (AL-JAMEEL, Abdul Jabbar Fadhıl Jameel, Metin Tutarlılığının Çıkarımsal Bir

    Modeli, Yüksek Lisans Tezi, Isparta, 2018)

    ÖZET

    Metin düzeninde, tutarlılığın rolüne geleneksel bakışı sorgulayarak bu araştırma,

    tutarlılığın çıkarımsal modeli vasıtasıyla gerçekleştirilen “araştırma-sonra-anlam-uyum”

    yöntemine bilişsel bir yaklaşım önermektedir. Önerilen yaklaşım varsaymaktadır.

    Tutarlılık sadece dilde (statik ürün olarak) var olan bir şey değildir; aynı zamanda dili

    kullanıcılarının zihninde var olan bir şeydir ki bunun mantığı (i) metindeki altta yatan

    (şifrelenmemiş) içerik ile dilsel olarak sunulan bilgiyi ayırmayı, (ii) bütünselliği,

    sayesinde bir cümlenin bütün çıkarımsal ilişkilerinin anlamı oluşturmak için birleştiği

    bir kavram olarak düşünmeyi, makul kılar. Bu bağlamda, herhangi bir dil öğesinin

    anlamı, anlamın şifrelenmiş ve şifrelenmemiş yönlerinin karşılıklı ilişkisinden oluşur ki

    bunun yorumlaması, sırasıyla, içerik açısından ve kasıtlı olarak uygun çıkarımlar

    oluşturmaya bağlıdır. Kabul edilen yaklaşımın gelişmesi için, çıkarımın farklı türlerinin,

    farklı iletişim türlerinde nasıl belirlendiğini ve nasıl “davrandığını” açığa kavuşturmak

    gerekmektedir. Çıkarım-temel-araştırma-sonra-anlam-uyumun dil kullanımının farklı

    düzeylerinde ne derece ilintili olduğu da bu araştırmanın temel ilgi odaklarındandır.

    Key Words: Metin Düzeni, Tutarlılık, Cümle içi-/Cümleler arası çıkarım temelli

    bütünlük, evrensel tutarlılık, çıkarım türleri, araştırma-sonra-anlam-birlik.

  • v

    (AL-JAMEEL, Abdul Jabbar Fadhil Jameel, An Inferential Model of Text

    Coherence, Master’s Thesis, Isparta, 2018)

    ABSTRACT

    Challenging the traditional view of the role of coherence in text organization, the

    research proposes a cognitive approach to “search- after- meaning- unity” procedure

    realized by means of an inferential model of coherence. The suggested approach

    assumes that coherence is not only something that exists in the language (as a static

    product) , but also something that exists in the brain of the language user, the logic of

    which makes it reasonable to : (i) distinguish between the underlying ( non - encoded)

    content and the linguistically represented information in the text, and (ii) consider

    meaning holism a concept which is the relevant unity of the non-encoded inferential

    relations and the physically expressed meaning of a sentence or an intersentential level.

    In this sense, a meaning of any language unit is composed of the interrelationship of the

    encoded and non - encoded aspects of meaning, the interpretation of which, in turn,

    depends on generating contextually and intentionally appropriate inferences, What the

    accepted approach necessitates in furtherance is to clarify how the different types of

    inference get determined in different genres of communication. To what extent the

    “inference-based-search-after-meaning- unity” is relevant at different levels of language

    use is also among the major concerns of the research.

    Key Words: text organization, coherence, intrasentential, intersentential, /

    inference - based coherence, global coherence, types of inferences, search - after –

    meaning – unity.

  • vi

    TABLE OF CONTENTS

    TEZ SAVUNMA SINAV TUTANAĞI .......................................................................... i

    YEMİN METNİ .............................................................................................................. ii

    ACKNOWLEDGEMENT ............................................................................................. iii

    ÖZET ............................................................................................................................... iv

    ABSTRACT ..................................................................................................................... v

    TABLE OF CONTENTS ............................................................................................... vi

    LIST OF FIGURES ..................................................................................................... viii

    LIST OF TABLES ......................................................................................................... ix

    LIST OF ABBREVIATIONS ........................................................................................ x

    INTRODUCTION ........................................................................................................... 1

    CHAPTER ONE

    1. COHERENCE AND TEXT ORGANIZATION

    1.1. FROM SENTENCE TO TEXT .............................................................................. 6

    1.2. COHERENCE AND COHESION ....................................................................... 11

    1.3. CRITICISM OF THE COHESION VIEW OF COHERENCE ........................... 13

    1.4. PRAGMATIC INFERENCE AND COHERENCE IN THE RELEVANCE

    THEORY ............................................................................................................. 18

    1.5. INFERENCE RESEARCH METHODS .............................................................. 24

    CHAPTER TWO

    2. THE TAXONOMY OF INFERENCES

    2.1. GENERAL PRINCIPLES: INFERENCE NECESSITATING FACTORS ......... 28

    2.2. THE TAXONOMY OF INFERENCE ................................................................. 33

    2.2.1. Logical- Semantic- Pragmatic Inferences ...................................................... 43

    2.2.2. Inference Favoring / Influencing Conditions ................................................. 61

    CHAPTER THREE

    3. INFERENCING IN DIFFERENT GENRES

    3.1. BACKGROUND .................................................................................................. 75

    3.2. INFERENCE - BASED COHERENCE SCIENCE ............................................. 76

  • vii

    3.3. INFERENCE –BASED COHERENCE IN A POLITICAL TEXT ..................... 83

    CONCLUSION .............................................................................................................. 92

    POSSIBLE PEDAGOGICAL IMPLICATIONS OF THE RESULTS AND

    FINDINGS OF THE RESEARCH .............................................................................. 94

    PERSPECTIVES .......................................................................................................... 95

    REFERENCES .............................................................................................................. 96

    ÖZGEÇMİŞ ................................................................................................................. 104

    CURRICULUM VITAE (CV) ................................................................................... 105

  • viii

    LIST OF FIGURES

    Figure 1. Propositions ..................................................................................................... 36

    Figure 2. “What Time is it? Simple Inferential Structure of Eleven Propositions .......... 37

    Figure 3. Adapted from Menmara and Magliano 2009................................................... 48

    Figure 4. Types of Inferences (Adapted from Rzayev: Unpublished Paper 3) ............... 48

  • ix

    LIST OF TABLES

    Table 1. Types of Inferences ........................................................................................... 54

    Table 2. Graesser et al.: Production Rules for Inferences (1994:381) ............................ 58

    Table 3. Distinction between Different Types of Inferences( Kispal 2008: 10) ............. 59

    Table 4. Inferences (Kispal 2008; 11) ............................................................................. 60

    Table 5. “Holland is flat” (Taken from Zvolensky 2015: 167). ...................................... 63

    Table 6. Types of Inference (adapted from Kispal 2008: 22) ......................................... 71

  • x

    LIST OF ABBREVIATIONS

    CDA : critical discourse analysis

    Cf : Confer (Latin) / See (English)

    CP : Cooperative Principle

    CV : Curriculum Vitae

    EFL : English as foreign language

    GDP : Gross Domestic Product

    LTM : long-Term Memory

    NCT : National Child birth Trust

    RT : Relevance Theory

    SCA : Socio-Cognitive approach

    SFG : Systemic Functional Grammar

  • 1

    INTRODUCTION

    Given that pragmatic inference has been richly studied in different fields of

    science, the main concern of the present research is to investigate the impact and role of

    inferencing (as a vehicle) in different genres in terms of distinguishing between the

    underlying (non-encoded) and the linguistically represented semantic content in

    language use. Also proceeding from the undeniable explanatory power and divergence

    of inferential mechanisms from the perspective of cognition and philosophy, as well as

    psychology, some linguistics claim that such kind of distinction (especially their

    proportional distribution) directly leads to understanding how text organization

    influences text comprehension (Brewer 1980: 229-233; Brewer and Lichtenstein 1982:

    476- 477; Bock and Brewer 1985: 55-60; Ohtsuka and Brewer 1988: 3-14. Brewer

    (1987: 187-197) which rests on the claim that discourse comprehension can hardly be

    isolated from the mental model of the world in discourse. And it is this framework

    within which the reader constructs his/her own mental model not breaking away from

    the writer’s original mental model (cf: Johnson laird 1983). As to the nature and

    working mechanisms of inference, both cognitive linguistics and the Relevance theory

    have given more than enough space to them. Even some inference researchers, such as

    Tendahl and Gibbs (2008) and Zhang Hui (2005) etc., have tried to combine these two

    approaches together to form a more comprehensive inference theory, since, for them,

    the relevance theory does not figure out the interconnectedness of inferential processes

    with coherence, which is of crucial importance for appropriate comprehension in detail.

    Thus, the paper will firstly make a brief summary of the studies on coherence (in terms

    of its contribution to text organization) and pragmatic inference under the existing

    theories, which will, in turn, lead us to the coherence providing power of inference.

    The contribution of different types of inference in comprehension of examples

    chosen from different genres is also among the prominent “missions’’ of the research.

    Accordingly, this research focuses on the impact of the underlying (i.e., non-

    encoded) content on the text, which, in turn, can be revealed by applying the inferential

    mechanisms suggested by the constructionist theory (Graesser et al 1994) and Rzayev

    (the unpublished paper), according to which, no author will provide full information in

  • 2

    the text structure and thus make the reader generate bridging inferences to recover the

    underlying context of the non- expressed events. E.g. consider the text borrowed from

    Noordman and Vonk (2012: 37): There were municipal elections yesterday. Because

    the majority of the lower town voted for the local party, there was a shift toward the left

    in the city council. But the right-wing party was not completely disappointed. It has

    expected to lose much more.

    But detecting any non-encoded meaning in the given text becomes possible via

    addressing the explicitly represented meaning of the text. Taking into account this

    principle, we can infer in the text above that the city council was more right oriented;

    that there is a contrast between the shift toward the left and not being completely

    disappointed, and, besides, there are other reasons (worse expectations) why the right-

    wing party was not so disappointed; and that ‘’it’’ refers to the right-wing party and not,

    for example, to the city council. All these pieces of information are not stated explicitly

    in the text, but, we, as readers, do understand them due to the so-called “inferences’’.

    Assuming the role of inferential semantics for sentences, Devitt (1996: 10)

    defines meaning holism as the view that the inferential relations a sentence contains

    together with the observable part constitute its whole meaning. To make it more precise,

    theory of making the meanings of sentences determines which inferences the speaker, or

    thinker, accepts and proceeds from. More than that, the relation that holds between two

    or more sentences is not only semantic. It, also, depends on the hearer. Given such non-

    semantic restricting facts, it is up to the theory to judge how the inferential properties in

    the “text-reader” interrelationship gets determined. That is, as Pagin assumes, the

    sentences involved must be assigned meaning to the effect that the accepted inferences

    come out as valid (2008; 219).

    Our approach: is both qualitative and partially quantitative in the sense that (i)

    the derivation of the speaker’s prior knowledge-based inferential meaning (s) can

    qualitatively differ from those of the hearer; (ii) Ignorance of the validity (i.e., truth

    fullness) of the inferential meaning can lead us to false or inappropriate conclusions. To

    put it otherwise, the qualitative approach assumes that the system of inference the

    essential feature of which is the preservation of truth is sound if and only if it never

    permits the inference of a false conclusion come from true premises; and (iii)

  • 3

    disagreement between the premises and the conclusion can create an ambiguity or

    misunderstanding. As for the partial quantitative nature of our approach, inferential

    meanings, as out-text components, (can) / (do) change the scope of the explicitly

    expressed meaning essentially, without destroying its central idea.

    We mainly take the stance of the Relevance theory which assumes that the gap

    between the encoded and non-encoded meanings of a definite description contributes to

    propositional content in a particular context which, in turn, is bridged by relevance-

    guided pragmatic inference (Carston and Powell 2008: 347). Also enhancing the

    Constructionist Theory’s basic principle “search after meaning” (Bartlett 1932; Berlyne

    1949; Spiro 1980; Stein and Jrabasso 1985; Graesser et al 1994), our approach implies

    the following critical assumptions:

    1. The reader, as an inferencer, constructs his / her own meaning of representation

    that arises as a result of his/her combining the explicitly expressed meaning with its

    deep level comprehension adequate to a definite referential situation;

    2. The inference constructs a coherent meaning representation at both local and

    global levels.

    3. Studying the question which kinds of semantics can be reasonably seen as

    engendered by inferences, we, as stated above, restrict ourselves to the question of the

    feasibility of seeing the meanings of the explicitly expressed representations in terms of

    their belonging to the classical logic in an inferential way.

    The Significance of the research: is due to the following gaps it aims to fill in:

    1. The research extensively investigates the inference-based discourse

    comprehension with respect to texts in different genres:, scientific texts and political

    texts;

    2. Given that little attention has been paid to (a) how local and global causal-

    consequence coherence relations are achieved at different levels, as well as (b) the role

    and potential of different types of inferences in the above-mentioned text types, the

    study focuses on the cognitive processes involved in comprehending the

  • 4

    interrelationships of the explicitly expressed and non-encoded meanings in different

    genres;

    3. Applying a new causal-consequence inference-Maker model (Rzayev,

    unpublished paper: 5) at intrasentential, intersentential and extratextual levels in

    Grammar, translation, semantics, literature, writing and other classes will enable

    students to find an adequate causal explanation for the integration of in-text and out-text

    meanings represented by different types of inference, due to which, causally activating

    the inferential meaning and connecting it with the text- base meaning, the student will

    extract the text as a set of inference-based integrated ideas:

    Structuring of the research: In accord with the aim and other specific features of

    the research paper, the “labour division ‘’ has been distributed as follows:

    The introduction part provides brief information about the basic aim, scope,

    approach, methodology, significance as well as the structural organization of the

    research.

    Chapter 1 is concerned with the inference-based coherence and text

    comprehension including a critical overview of the existing theories on coherence and

    inferential models of communication. “Methods in inferential research of text

    comprehension” part is also amongst the major concerns of Chapter 1.

    The second Chapter mainly deals with the taxonomy of inferences as well as the

    inference influencing factors.

    Chapter 3 is concerned with both theoretical and empirical aspects of inferencing

    in different genres of language use. The ways different types of inference provide

    coherence in different genres constitute the major task of Chapter3.

    Conclusion “mirrors’’ the results and findings of the research while

    “Pedagogical implications’’ part focuses on the implementation of the principle findings

    of the study in improving teaching of English as a foreign language at the universities of

    Turkish Republic.

  • 5

    The perspectives part emphasizes those aspects of coherence and inference

    which are still open to discussion in terms of the ways of benefiting from the realization

    of causal and predicting inferences in the processing and comprehension of text types in

    the unexplored genres as well as in spontaneous spoken discourse. References include

    two types of sources: the directly used/cited literature and the supporting citation

    references.

  • 6

    CHAPTER ONE

    1. COHERENCE AND TEXT ORGANIZATION

    1.1. FROM SENTENCE TO TEXT

    Text organization, as a complex process, reflects in itself ( a ) constructing a

    cognitive representation of the content conveyed by the text ( as a whole) (Meyer 1984;

    Murray 1995; Jones 2012; Widdowson 1973; Harris 1952, 1964; Labov 1972; Bee

    2010;Baker 2005; Bhatia 1995; Bibereta 2007) as well as (b) the pragmatic-dialectical

    contributions of the different syntactic-semantic devices within the text (Kitsch and Van

    Dijk 1978;Sanders and Norman 2000;Liu 2009;Martin 1992;Schegloff

    2007;Schiffrinetal 2004;Scol-Lon 2007; Van Dijk 2008; Widdowson 2007;Martin

    1985;1992).

    People who approach texts from the point of its organization and functioning

    usually focus on four different perspectives based on four different ideas about what

    text is. Some have taken a formal approach to text, seeing it simply as language unit

    above the level of a clause or sentence (Harris; 1952:1). What Harris aimed at here was

    to throw light on how sentences are put together to form texts (Harris 1952 in Jones

    2012:36).

    The idea that texts could be analyzed compositionally was actually very popular

    in the early and mid-twentieth century (under the influence of the dominant structural

    approach to language), even before Harris introduced the term “discourse analysis”.

    E.g., a group of Russian Formalist, tried to apply the principles of the grammar

    of sentences to analyzing stories and novels. Vladimir Propp (1986) was well-known for

    coming up with ‘’grammar of stories” in studying Russian folk tales. The “distributional

    analysis” proposed by Harris, in fact, was not much different from the widely used

    grammatical analysis of a separately taken sentence. The idea was to identify particular

    linguistic features and determine how they occur in texts relative to other units, or “in

    the same environment” with them.

  • 7

    This approach gives rise to questions which are crucial not only from the point

    of information structuring within the frame of communication development, but also in

    terms of behaviour of text structures in different functional styles as well as in certain

    kinds of social practices: How could such knowledge explain whether “old” refers to

    only “men” or it modifies both “men” and” women” in the sentence, or how can the

    logic of sentence-level grammar remain the same within longer stretches of text without

    causing unavoidable complexities in comprehension (problem1) and how does language

    correlate with” culture” (i.e. linguistic and non-linguistic behaviour)? (Harris 1952/

    1964: 356).

    For Widdowson (1973-in Jones 2012: 137), Harris simply establishes formal

    patterns without any reference to meaning, what contradicts the basic function of

    language in terms of its communicating a definite meaning. For Halliday, the

    distributional analysis also demonstrates how the formal properties of different pieces of

    language correlate with the social situations in which they occur.

    Widdowson, unlike Harris, argues that the analysis of discourse or text must

    reveal not only how texts are put together but also how people use language to perform

    social actions. Sharing this argument, William Labov, advanced the idea that “the object

    of linguistics must ultimately be the instrument of communication used by the speech

    community, and if we are not talking about that language, there is something trivial in

    our proceedings”( Labov 1972: 187) .

    Jones (2012: 138), using the term “discourse analysis” also refers to the way

    linguistic elements are put to communicative use in the performing of social action,

    whose purpose is to discover what sentences count as utterances and what it is that

    provides a discourse with its coherence “as a piece of communication”. Widdowson,

    contrasting Harris’s formal view of discourse with how language performs particular

    communicative acts, associates the perspective of Harris more with the study of

    cohesion and the perspective of Labov(discourse as language in use) more with the

    study of coherence of discourse “ as a piece of communication”.

    For J. P. Gee (2010: 28-29). “People build identities and activities not just

    through language, they also use the resources of language to project themselves as a

  • 8

    certain kind of person (in accord with the assumption “language is indexical)”, in

    different circumstances (to protect your “social face”, etc). Thus, you behave as a

    different identity at a formal dinner party than you do at the family dinner table and,

    though these are both dinner, they are nonetheless different practices or activities

    (different “games”/ “roles)”.

    As Bhatia et al (2008:1) state, having undergone a “discursive turn”, social

    sciences have become increasingly interested in the creation of the reality that surrounds

    us. This interest, in turn, has given rise to new theories and methods for the study of

    language use and its role in human society. Being of interdisciplinary nature, discourse

    analysis has acquired the status, stability, significance and integrity of a well-established

    discipline. Unlike the 1960s, linguistics has extended its boundaries on an account of

    the limits of individual sentences, shifting its focus to the meaning constructed and

    interpreted specifically under the impact of particular social contexts.

    This definition stems from two relevant features: (a) language (in use) can be

    analyzed both on the level of (a) sentence/clause, and (b) on the level of the text.

    Accordingly, language ought to be analyzed not as an abstract set of rules, but as a

    dynamic tool for social action. Although early conceptualizations of text organization

    (e.g. the structuralist / formal approaches) were seen as offshoot of language in use,

    their present versions have moved to a focus on “ language in use,” benefiting from

    sociology, psychology, semiotics, communication studies, rhetoric as well as disciplines

    such as business and accountancy, organizational studies, law and information

    technology, to name only a few. Due to this enrichment, the use of language in a variety

    of institutional, academic, workplace and professional settings has become more

    rational and efficient.

    The roots of the view of text in terms of” language in use” are best traced to the

    work of Wittgenstein (1951 / 1972), for whom it is language through which people

    construct not only “forms of life”, but also communicate to others and their

    surroundings Bhatia et al 2008: 2). Less than two decades later with the publication of

    Austin’s 1962 classic. ‘How to do things with words’’; the notion that the study of text

    should involve more than just its structure but also the way it is used and the way

    structure shapes and gives rise to it became more prominent. Later, Foucault and

  • 9

    Derrida, though diverging considerably from the tradition of Austin, also made

    language, and in particular “discourse”/ text central to their understanding of social

    practice. Social scientific disciplines particularly concerned with social practice also

    began to recognize the centrality of language in much of what they were studying.

    Anthropologist Gregory Bateson and psychiatrist Jurgen Ruesch (Ruesch and Bateson

    1951) argued that social and psychological phenomena cannot be separated from the

    matrix of communication in which they occur. They were followed by a host of social

    and behavioural scientists, among them Goff-man (1959) and Garfinkel (1967), who,

    focused on the role of language in social behaviour and social formations. By the 1970s,

    psychology, sociology, and anthropology had all taken decidedly ‘’discursive’’ turn

    influenced not just by the structuralist linguistics of de Saussure, but also by a new

    group/generation of linguists who were becoming more and more aware of the

    relationship of language to social actions and to the socio-cultural worlds of those who

    use it.

    In America this concern had given rise to the ‘’linguistic relativism’’ of Edward

    Sapir and Benjamin Whorf while in Europe this new concern in use was first

    exemplified by the Prague linguistic circle( also called Prague functional school)

    representatives whose major “purpose” was to identify the potential of language units

    (no matter to what levels they belonged) from the point of their contribution to language

    use in the process of communication in different contextual situations for different

    purposes. Proceeding from the function of language in the social structure” (1973: 65),

    Halliday and Hasan named their approach “Systemic Functional Grammar (SFG)”

    which has had a profound influence on many contemporary schools of discourse/text

    linguistics, including Critical Discourse Analysis (CDA), Conversation Analysis and

    Multimodal Discourse Analysis.

    The analytical tools of the Prague functional school ( the oppositional analysis,

    the thematic progression and its types in accord with the purposive and contextual

    aspects of communication) and Halliday’s grammar as well as Functional grammar

    established by Dik, have been found rather productive in tracking participants in

    different basic sentence types, logical as well as communicative relations, processes,

  • 10

    qualities and evaluations of these by language users as they develop throughout a given

    text or across group of texts (cf. Martin and Rose2003) .

    These major discourse/text analytical approaches, to varying degrees, represent a

    concern for language use in the social world and focus on widely differing aspects of its

    use and often define the adaptability of communication (i.e., text processing and

    comprehension) to/from the immediate conversational context to the larger political,

    social or economic context...

    As the title of this research suggests, we are not interested in surveying all the

    available literature that has been done using these various approaches. In doing so we

    hope not only to understand what is specific about these approaches, but also to identify

    where future possibilities for convergence and interdisciplinary are opening up.

    However, it is first necessary to understand the main questions upon which these

    approaches diverge and the different roads they have taken from common intellectual

    roots. It is doubtless that those different approaches diverge on two of the most basic

    issues in this formulation: (a) the question what text is, and (b) how the affects text

    organization. These differences can be seen (in one possible sense) as different factors

    contribution of multiple disciplines to the development of text /discourse studies. E.g.

    sociologists and anthropologists have viewed the use of language as a function of the

    (social-cultural) context, whereas linguistics is mainly concerned with focusing on text

    structure with context in the background.

    In contrast to restricting our approach with the requirements of this or that

    trend, we think it reasonable to consider and benefit from where they might be going by

    (enriching each other) and how their trajectories (both horizontally and vertically) might

    converge to elaborate the findings as “ingredients” of the whole concerned with the

    study of text in terms of its coherent constituents as opposed to randomly used

    sentences which would not be considered a text (Carrell 1982: 479).Many researchers

    (as mentioned earlier) have been working hard trying to understand the fundamental

    properties of text and some theoretical accounts of them have been proposed. Often

    these accounts are connected with textual analysis techniques which parallel sentence

    analysis techniques. These approaches are even sometimes called” text grammars”.

  • 11

    Among others, in this sense, have been the American structuralist Charles Fries (1952),

    the first American transformationist Zelling Harris (1952 / 1970), the tagmemicists

    Kenneth Pike (1967) and Robert Longacre (1968, 1972), as well as a propositional

    analyst Walter Kintsch (1974), macrostructulist Teun Van Dijk (1972, 1977), and story

    grammarians Mandler and Johnson (1977), Stein and Glenn (1976), Thorndyke (1977)

    and Rumelhart (1975). The most influential of the widely used textual analysis

    techniques, however, in applied linguistics, has been the approach of Michael Halliday

    and Rugaya Hasan (1976), also known as cohesion theory.

    The reason of the organization of this part of chapter 1 is threefold. The first

    section of the paper reviews Halliday and Hasan’s (1976) cohesion concept in terms of

    textual coherence, while the second section is concerned with criticisms of the cohesive

    view of coherence and relevance.

    1.2. COHERENCE AND COHESION

    In spite of its being one of the central notions in the study of text, the concept of

    coherence has many different and often incompatible definitions and connotations. For

    text linguists or psycholinguists, coherence is predominantly matter of semantics and

    domain knowledge, while various brands of speech act and dialogue analysis describe

    coherence in terms of intentions and context dependencies.

    We will argue in this section that cohesion can hardly be taken as an attribute of

    the “connected and referential continuity”of text. We begin with the role of

    connectedness between /among the sentences of a text and how it can be provided in

    languages use.

    As some scholars (Sanders and Maat 2006: 591) point out, discourse is in no

    way a random set of utterances: it should show connectedness. Accordingly, a central

    objective of any linguistic work is to characterize this connectedness. Linguists have

    traditionally approached this problem in terms of presence of observable language

    devices and structures. For instance, Strube and Hann (1999: 309) argue that

    establishing referential coherence in discourse can be characterized as the presence of a

    proper antecedent of a given anaphoric expression in the current or the preceding

    utterances and the rendering of both as semantically identical. This task, as the

  • 12

    interpretation suggests, can also put general constraints on the grammatical

    compatibility of the expressions involved (cf: Haddock1987; Alshawi 1992). Linguists

    (from different approaches) have also made considerable efforts to identify the

    emerging and contribution of the scope and “behaviour” of intersentential anaphora

    within the representation theory (Kamp and Reyle 1993)or for intersentential anaphora

    within the Binding theory (Chomsky 1981), or within the context of the discourse

    representation theory (Kamp and Reyle 1993). Unfortunately, these approaches fail to

    differentiate coherence and cohesion, as a consequence of which, referentially

    ambiguous interpretations have to appear when several alternatives fulfil all the required

    syntactic and semantic constraints. It proves that syntactic and semantic criteria

    constitute necessary but by no means sufficient conditions for identifying cohesion and

    coherence in terms of text processing and comprehension activities.

    For Rider et al (2011: 455), cohesion as a set of relations combines together

    structures and lexical items to form a text. Pronouns like” she,” for example,

    conjunctions like “but|, and linking adverbs like “therefore” usually perform this or that

    cohesive role in a text. Rhetorical patterns, as Meyer (1985:21) states, apply to Macro-

    propositional semantic relations, which include “collection”, “causation,” response’’

    comparison” and “description’ also combining in a wide variety of ways to provide

    possibilities for the subordinate relational structures of exposition.

    For Recento (1987: 68 -.69), cultural/values/attitudes approach is based on

    analysis of paragraph orderings and definitely plays a role in ranking thematic material

    in cases in which there is thematic ambiguity or more than one macro-proposition in

    text.

    For Rzayev et al (2016: 236), connectedness provided by both cohesiveness and

    coherence functions to make sure if: (a) all the parts of composition are closely related

    to each other in explaining one common idea and (b) if they are relevant and ranged in a

    clear order, ask and answer the appropriate questions. For Carrell (1982: 480), Halliday

    and Hasan’s (1976) treatment of cohesion as a device contributing to coherence,

    overlaps with the tradition of text grammars which usually treat text primarily as only a

    structure-based linguistic phenomenon. For Halliday and Hasan, “cohesion does not

    concern what a text means; it concerns how the text is constructed as a semantic

  • 13

    edifice” (1976: 26). Assuming that cohesion exists within text between a presupposed

    item and a presupposing item, Halliday and Hasan (1976:4) also suggest that: ‘’The

    concept of ties makes it possible to analyze a text in terms of its cohesive properties and

    give a systematic account of its pattern of texture” (ibid: 4).

    However, as Blakemore (1998:85) states, meaning relations are not always

    realized explicitly for a discourse to have coherence.e.g. both the causal connection in

    “he was tired and so he went to bed” and the temporal connection in” he wrote a letter

    and then went to bed” could have been conveyed implicitly i.e., without “so” in the 1st

    and “then” in the 2nd case.

    Moreover, even a cohesive tie can combines two sentences/clauses to go beyond

    the encoded linguistic resources to recover an interpretation. For example “he” could in

    principle refer to either John or Bill in the sentences borrowed from Hobbs (1979):

    John can open Bill’s safe. He knows the combination.

    This, in fact, suggests that linguistic connectivity alone can hardly provide

    content connectivity. Hobbs (1978), for example, argues that the coherence of a text or

    discourse can be defined in terms of a set of structural binary relations between its

    segments, which, in turn, depend on their propositional content.

    A speaker who wishes to be understood must ensure that his utterance stands in

    one of these relations to the preceding text, and a hearer who wishes to understand the

    utterance must recognize which particular relation it bears to the preceding text (Blake -

    more 1998: 85

    1.3. CRITICISM OF THE COHESION VIEW OF COHERENCE

    The other sources of criticism concerning the cohesion view of textual coherence

    come from even those researchers who share several different perspectives (Brown and

    Attardo 2008; Feathers 1981; Morgan and Sellner 1980, Tierney and Mosenthal 1981;

    Free body and Anderson 1981; Steffensen 1981; Carrell 1982).

  • 14

    E.g. Karen Feathers (1981) assumes that it would be more rational and

    reasonable to analyze a text into its propositional units and then look for cohesive ties

    between the propositions, rather than the surface structures.

    E g., having examined the relationship of cohesion to coherence, the report

    below results consistent with Morgan and Sellner’s theory - based criticism of Halliday

    and Hasan.

    Tierney and Mosenthal (1981) experimented this by correlating the propositional

    use of cohesive ties by 12th grade students in written essays with holistic coherence

    rankings given by their instructors. The students, elicited under controlled conditions,

    with several other variables of interest in the study (e.g., Topic and familiarity)

    controlled that” topic” or” content”, as Morgan and Sellner argued, appears to affect the

    options a writer has for using cohesive items. Summarizing the findings of the

    experiment, Tierney and Mosenthal conclude: “What was found is that cohesive ties are

    pervasive in text and are patterned across topics. But ties are pervasive almost by

    definition since reference, conjunction, and lexical cohesion include a large proportion

    of any text… such pervasiveness severely diminishes the usefulness of the cohesion

    concept as an index of coherence… This study argues against using cohesion analysis as

    an index or predictor of a text’s coherence. There appears to be no causal relationship

    between proportional measures of cohesive ties and coherence rankings with a topic.

    The present study indicates that a cohesion index is causally unrelated to a text’s

    coherence” (Tierney and Mosenthal 1981: 24-25).

    Freebody and Anderson (1981), empirically studying the effect of three different

    levels of cohesion on reader’s comprehension, looked at the effects of vocabulary

    difficulty, which had a dramatic effect on comprehension, but the amount of cohesion

    did not. Hagerup Nelson in (1977) indicates that lack of connectives does not seriously

    damage comprehension because readers are usually able to make bridging

    inferences”(1981:19).

    To reveal the interactive effects of both cohesive ties and cultural background

    knowledge on reader’s processing of short prose texts, Steffensen (1981) reads a

  • 15

    comparable passage from both their native culture and from a foreign culture, and asked

    them to write their recalls of the passages.

    Analyzing the results, Steffensen found that causal cohesive elements were

    recalled better by readers from the passage of their own native culture than from the

    passage of the foreign culture. These findings suggest that when there is mismatch in

    cultural background knowledge between the reader and that assumed by the text, there

    will be a loss of textual coherence. For Steffensen, textual coherence can be fully

    realized only when a reader appropriately identifies the schema underlying a text. In

    other words, if a reader does not have or fails to access the appropriate background

    schema underlying the text, all the cohesive ties in the world won’t help that text cohere

    for that reader.

    For Brown and Attardo, (2008) coherence is the overall meaning of a text.

    Accordingly, you may think of it as its” point” or” main” idea”; or as the part of its

    meaning that makes it all fit together. Happening at the semantic level, coherence may,

    but does have to be, explicitly expressed in the text itself. Besides by cohesion,

    coherence may be established by any of the following means:

    1. The setting up and fulfilment of expectations in the text. e.g.; we start talking

    about a couple engaged to be married and then we describe getting married;

    2. The cooperative principle, i.e., by implicature, for example, one asks for walking

    directions to a given address, only to be told that the address is five miles away; the

    answer assumes the implicature that the question is being asked with the purpose of

    going to the address.

    3. Reasoning , I.e., by inference, for example, if we say John wants to the

    swimming pool and he asks to borrow Mary’s car, we infer that he is doing so in order

    to reach the swimming pool;

    4. By the activation of our knowledge of a common situation .e.g., if we are

    discussing a trip to the restaurant, we will expect to find mention of food, whereas if we

    are discussing “the stars in Hollywood,” the word “star” is most likely to be interpreted

    as” famous people” rather than” heavenly bodies” whereas in with a telescope , you can

  • 16

    see many stars in the sky” the context selects the other interpretations as more likely

    (Brown and Attardo 2008: 52,68).

    The presence in a text of cohesive devices, as they claim, does not guarantee that

    it is also coherent, although usually coherence and cohesion go together. To show that

    cohesion can neither be sufficient nor necessary to provide consider the following two

    paragraphs:

    (A) John likes to swim. Mary is fond of sky diving. Ann is a progolfer what

    athletic children have.

    (B) John likes to swim. It is a very good sport, from an exercising point of view.

    Exercise is a good way to lose weight. Weight loss is the number one reason for dieting.

    In (A) there are no cohesive ties (unless you count the fact that John, Mary, and

    Ann are potentially children’s names) and yet coherence is easily achieved by invoking

    the frame for ‘family’’ which tells us that one may have three children.

    In (B) there are cohesive ties between each sentence and the following one, yet

    the paragraph fails to be coherent because there is no unifying theme. (Brown and

    aflardo 2008: 53-54).

    Proceeding from the assumption that the cohesion approach to connectedness is

    inadequate, Sanders and Maat argue: “the dominant view has come to be that

    connectedness of discourse is a characteristic of the mental representation of the text

    rather than of the text itself. The connectedness thus conceived is often called

    coherence.” (Sanders and Maat 2006: 592)

    Generally speaking, they distinguish two respects in which texts can cohere:

    1. Referential coherence: smaller linguistic units (often nominal groups) may relate

    to the same mental referent, i.e., discourse anaphor

    2. Relational coherence: text segments (most often conceived of as clauses) are

    connected by coherence relations like cause -consequence between them.

    What they emphasize is: “we need to realize that coherence phenomena may be

    of a cognitive nature, but their reconstruction is often based on linguistic signals in the

  • 17

    text. Both coherence phenomena under consideration-referential and relational

    coherence-have clear linguistic indicators” (Sanders and Maat 2006:592). They also

    claim: “For referential coherence these are devices such as pronouns and demonstratives

    while for relational coherence these are connectives and (other) lexical markers of

    relation, such as cue phrases and signalling phrases. A major research issue in this case

    is the relation between the linguistic surface code (what Givon 1995 calls “grammar as a

    processing instructor)” and aspect of the discourse representation” (ibid: 593).

    Referential coherence relation can be illustrated by different referential devices

    which correspond to different degrees of activation for the referent in question. For

    instance, a discourse topic may be mentioned in the first sentence but once the referent

    has been identified, pronominal forms suffice. This can hardly be taken as a

    coincidence. Many linguistic have noted this regularity (e.g. Ariel 1990 Givon 1992;

    Chafe 1994).

    Ariel (1990, 2001), for instance, in her “Accessibility Theory” argues that high

    accessibility markers use little linguistic material and signal the default choice of

    continued activation. By contrast, low accessibility markers contain more linguistic

    materials and signal the introduction of a new referent’s.

    If to consider coherence relations in terms of their functioning, we are to state

    that they are also termed as rhetorical relations” (Mann and Thompson 1986; 1988;

    1992) or” clause relations” which constitute” discourse patterns” at a higher text level

    (Hoey 1983). Cohesive relations, being meaning relations, connect two or more

    sentences, cohesion’s contribution to this property of discourse implying that sentences

    mean more together than apart (Hoey 1996:13).

    The other defining characteristic for these relations is that they provide more

    information than is provided by the sum of the segments taken in isolation (Sanders and

    Maat 2006:593). For Rzayev et al (2016:523), each discourse element is classified

    according to the communicative function it permits in relation to the discourse as a

    whole.

    The patterning of relations and elements in a discourse is described as the

    discourse Macro- pattern, that is, the overall composition of discourse is determined by

  • 18

    the order in which each of its elements occurs. For example, as Crombie (1985; 58-59)

    and Maurer (1999: 182-183) argue, a discourse consisting of four elements (situation,

    problem, solution and evaluation) is usually constructed on the three binary discourse

    relations (situation-problem; problem-solution and solution-evaluation).

    For Sanders and Matt (2006:593), although relations like cause-consequence are

    conceptual, they can but need not be expressed explicitly.

    Recently, a significant part of research on coherence relations, focusing on the

    question of how different sets of relations should be organized (Hoey 1990: Knott and

    Dale 1994; Sanders etal 1992), has started to define the “relations among relations”

    relying on the intuition that some coherent relations are more alike than others. For

    instance, the relations in (A),(B) and (C), all express a certain type of causality; they

    express relations of cause-consequence volitional result: A-argument-claim; conclusion

    (B) and speech act causality (c): “This is boring watching this stupid bird all the time. I

    propose we go home now”. The relations expressed in (D) and (E), however, do not

    express causal, but rather additive relations. Furthermore, a negative relation is

    expressed in (D). All other examples express positive relations and (E) expresses an

    enumeration relation”.

    A. The eagle was looking for prey. The bird was soaring in the air for hours.

    B.The bird has been soaring in the air for hours now. It must be an eagle.

    C. The eagle has been soaring in the air for hours now. Let’s finally go home.

    D. The eagle was soaring in the air for hours. Yesterday we did not see it all day.

    E. The eagle was soaring in the air for hours. There was a peregrine falcon in the

    area, too.

    1.4. PRAGMATIC INFERENCE AND COHERENCE IN THE

    RELEVANCE THEORY

    In everyday communication, the addressee often does not take the time to

    exactly capture the speaker’s intended meaning, since his / her role includes both taking

    in what is said (communicative meaning or informative intention) and inferring the

  • 19

    speaker’s intended meaning (i.e., communicative intention), for what the addressee is to

    understand to infer what is behind the addressor’s utterance, which is usually called

    pragmatic/ inference. What stimulates the inferential activity is the fact that there is a

    lack of information and the reasons of such kind of asymmetry between the

    linguistically encoded meaning and the communicatively intended meaning are rather

    various:

    (1) The economical and asymmetric principle of language use;

    (2) The speaker’s “cloudy” use of language (intentional or unintentional)

    (3) the listener’s incompetence in comprehending, inferring and so on, in which

    situation the addressee may still fail to get the real meaning (either both the encoded and

    intended meanings or only the intended meaning). Here, only the first reason will be

    discussed.

    The informative intention of an utterance is much easier to comprehend than the

    communicative intention behind it for the latter needs more “information” and the

    language (i.e., the economically expressed meaning) is far from enough. Relevance

    theory mainly is concerned with the ways how people get information from an utterance

    by inferring (deductively mostly) it in a cognitive context (Heziran 2006). Context

    being definitely something indispensable to pragmatic inference (as well as to the whole

    communication), the relevance theory takes communication as a process of inferring in

    a specific context. Although the existing opinions are widely divided in terms of the

    definition of the term context” (and it is still an open question), relevance theory has

    made great progress by coming up with the notion of “cognitive context”, which differs

    from the traditional context.

    Meanwhile, different scholars focus specifically on different aspects of the

    cognitive context, justifying this paradox with an assumption like the following:

    Context is neither objective entity nor something predetermined; it is something coming

    out of dynamic inferring process (Cai Yun 1997), which, for example, in relevance

    theory, is also called cognitive assumptions, including the logical information,

    encyclopaedic knowledge, and lexical representation; it is a’’ psychological construct ;”

    it is part of the assumptions about the world; it is based on the real world and more than

  • 20

    that as well. The process of contextual assumption must be inferred while the inference

    is an intellectual enquiry process. In Ostensive-inferential communicative model, the

    speaker shows the listener his informative and communicative intention by means of

    ostensive behaviours, thus providing necessary grounds of judgment of inference.

    Meanwhile the addressee makes inference and forms contextual assumptions according

    to the addressor’s ostensive behaviours ,the final aim is to reach optimal relevance and

    communicate successfully, that is, to get adequate effect for no unjustifiable effort. So

    to infer is to search relevance. The inferential mechanism of relevance theory is closely

    connected with the concepts of maximal relevance and cognitive context. For example,

    in the beginning of an American film titled” pretty woman” Edward quarrelled with his

    girlfriend on the phone and she complained:” I speak to your secretary more than I

    speak to you”. Then Edward met his former girl friend Susan and the following

    conversation takes place:

    Edward: when you and I were dating, did you speak to my secretary more than

    you spoke to me?

    Susan: She was one of my bride maids.

    Prescriptively, Susan’s answer is irrelevant with Edward’s question, because

    Edward expects an answer with “yes” or” no”; but Susan doesn’t answer Edward’s

    question directly. To understand the implied meaning, some relevance must be

    established and from what Susan has said, the following assumptions may be formed in

    Edward’s mind:

    (a) If one wants to be Susan’s bridesmaid, one must be very familiar with her.

    (Inference: Edward’s secretary was very familiar with Susan.)

    (b) Edward’s secretary became Susan’s bridesmaid, so she must be Susan’s good

    friend.

    (c). It is true that Susan spoke to Edward’s secretary more than he spoke to her.

    In this way Edward, understanding what Susan means, can achieve the optimal

    relevance. Thus, Sperber and Wilson’s ostensive and inferring model is impossible

  • 21

    without inferring (Xu Shenghuan 2007: 3). It can also be said that inference is the core

    of the relevance theory. E.g., the communicative intention of the addressor is not

    exactly encoded in literal language; nor can it be accessed through sentence phrase.

    Accordingly, comprehending as a non-demonstrative inferential process can be divided

    into two stages: assumptions formation and testing (Ran Yongping 2002: 52).

    Being a universal necessity (in terms of economic and rational use of language),

    inferencing is, in particular, of crucial importance in indirect speech acts. The following

    example depicts an episode from a bus crash.

    A: so taking bus is not safe.

    B: 130 is a plane, which you can take.

    As seen from B’s answer,” 130 is a plane” is a typical metaphor in the context of

    A’s utterance. But we realize that it is a metaphor. “130” is, in fact, the number of a bus,

    which is only known to people who live in this district where bus no.”130” runs. And

    this is part of shared knowledge manifested between A and B. Furthermore, each of

    them knows that planes have wings and are of super speed; with this in mind, it can be

    inferred that B is implying that bus no.130 is too speed and not safe enough to take,

    thus” do not take it” is what he really intends to say.

    While cognitive content, as a dynamic process is applied to test assumptions,

    inferring process, being cognitively ever-changing, functions as a “bridge” between the

    linguistically encoded meaning and the non coded meaning, which, for Rzayev

    (unpublished paper: 2) consists of two constituents: 1) the decoded informative

    meaning, and (2) the intentional meaning. As also seen from the example above, only

    “do not take bus no. 130” is the intentional constituent of the inferred information which

    in turn becomes relevant only after inferencing the non coded informative information

    (“bus no. 130”is too speedy and not safe enough to take”). This cognitively ever-

    changing nature might play an essential role in both processing and comprehension of

    communication.

    For Hofweber (2010: 1-26), there is also an undeniably strong unity between the

    intended meaning and inferential role in our language. And the expressions, to exhibit

    this connection are inherently related to other sentences in a certain way (Hofweber

  • 22

    2010:2). But this assumption does not require that speaker is disposed to prefer this but

    not the other way, or that they do it to Show that inferential way is best in this case. The

    expressions in question, including the predicate logic (the semantic level), can have an

    inferential role which should in no way break away from the intended meaning. To see

    how this happens, what kind of inferential role the connectives in connection with the

    truth predicates are supposed to have.

    E.g. a crucial difference between the inferential behaviour of expressions like

    “and” and “dog” is evident to everybody and in all cases “dog” figures in all kinds of

    valid inferences (whether literal or metaphorical), which, as the truth of the premises,

    guarantee the truth of the conclusion. “Fido” is a dog” implies” Fido is an animal”. But

    “and’’ in contrast is characterized not by an inferential relation to various other

    sentences with or without “and” in them. The inferential role of “and” is captured by a

    simple Schema: And in this Schema everything is left Schematic except for the “and”.

    This is paradigmatically done (for economizing purposes) with the natural deduction,

    inclusion and/or elimination rules.

    Yus (2006:1) goes still further in arguing that “… the development of

    pragmatics, especially cognition-cantered pragmatics, has made us aware that what is

    coded in human communication (spoken utterances, written texts, nonverbal behaviour,

    images) highly underdetermines the information that the Sender actually intends to

    communicate, that is, coded messages are far less informative than the thoughts that the

    Sender really wants to communicate with them”.

    For Yus and other proponents of the Relevance theory, within the more

    appropriate “inferential model” of communication, although coding is necessary in

    communication (we do need utterances, written texts, etc. to transmit information to one

    another), it is inference that plays a major role in transforming the schematically coded

    messages into fully satisfactory interpretations.

    If to assume that the point of a philosophy of presence is to accomplish certain

    things not only in the text but beyond the text (Said 1978: 681), then we are forced to

    say that the events in the text demand an active, questioning interpretation which must

  • 23

    be approached on two levels: first, from within the text and second, from the

    perspective of the reader (Gooze 1983: 337).

    The inferentialism beginning with the work of Paul Grice (1989: part 1) treats

    utterances / evidences about the communicator’s intentions and assumes that “the

    borderline between decoding and inference coincides with the borderline between

    explicit and implicit communication, so that the proposition and attitude expressed

    would be recovered purely by decoding and the role of inference would be restricted to

    the calculation of implicature’ (Wilson 1998:5), what, in fact, overlaps with the Grician

    cooperative principle and maxims for whose opinion inferential intention is

    characteristic for implicit communication. If, for example, your friend asks you if you

    think her new boyfriend is good looking, but you do not think so, you might express

    your opinion saying “He has a lovely personality, “violating the maxim of relevance

    (her question was not about his personality, or you might say something rather vague

    meaning. Showing that you do not think he is very good looking thus avoiding saying

    this explicitly (violating the maxim of manner). The obvious questions are: why do

    people do this? Why don’t they simply communicate what they mean directly?

    For Jones (2012; 54), one reason is that implicative way of language use allows

    us to be politer or to avoid hurting someone’s feelings. We also use implicature not to

    make ourselves too accountable for what we have said-in other words, to say something

    without “really saying” it. Most scholars distinguish two kinds of inferences. The first

    kind is an inference used to generate new information. For example, from the sentence

    “John selected his strantson because there was much mud” the reader can infer that

    apparently strantson is a material or brand that has advantages if there is much mud.

    Given that the reader is not familiar with strantson, the reader cannot know that the

    conjunction” because’ is correctly used. However, by assuming that the sentences make

    sense, the reader can derive the inference as new information. The second kind of

    inference is a knowledge activating one. Examples from the text (about municipal

    elections) are that “it” refers to the right - wing party, and that, given the contrasted

    relation indicated by “but”, a shift toward an opponent party is in general a reason for

    disappointment. Although this kind of inference in general is not recognized in

    everyday language, most of the psycho-linguistic research can hardly neglect it.

  • 24

    The other unchallenged thing (especially concerning the second type of

    inference) is that in understanding a text/discourse, the number and variety of derivable

    inferences seem to be almost unlimited. Yet, as the comprehension is accomplished too

    quickly for many time-consuming inferences to be made, the first issue is how to

    account for the control of inferences: which inferences are made and which ones are not

    made? The second issue concerning the inference as a process is; what is the process

    that allows us to make an inference? How does the inferred information get activated

    and deactivated? How does the information in the text interact with the reader’s

    knowledge? Although until recently, inferences were considered as a rather isolated

    phenomenon, our research considers inferences as one of the components of the

    comprehension process. The newly developed methods to measure brain activity also

    lead to a better understanding of the relation between inferences and the brain.

    1.5. INFERENCE RESEARCH METHODS

    When and what kind of inference is generated due to the assumption claiming

    that the inferences stemming from the mental text representation, are similar to

    information that is expressed explicitly by the text. Therefore, they require fulfilment of

    the tasks such as reproduction, recognition, and verification. In a production task, the

    reader reproduces the information that was not stated explicitly in the text. In a

    recognition task, the reader has to judge to what extent particular words or sentences

    contribute to inference generation; the necessity of the recognition task is similar to that

    of the reproduction task.

    Many people who had studied the sentences “John was trying to fix the bird-

    house. He was pounding the nail when his father came out to watch him and to help him

    to do the work” incorrectly recognized the sentence “John was using the hammer to fix

    the bird house when his father came out to watch him and to help him do the work”.

    This indicates that an instrumental inference (hammer) was made (Bransford and John-

    son 1973). But reproduction and recognition tasks, indicating whether the inferences are

    made or not, but in general cannot distinguish whether the inferences are made during

    reading (at encoding) or during the measurement (at retrieval). In verification tasks, the

    generated inferences verify whether the content of a sentence is true or false with

    respect to the content of the text. Off-line or after-reading methods, used in combination

  • 25

    with online or during reading methods answer the question of when inferences are

    made.

    Online methods, aiming to detect the ongoing inference process immediately,

    improve comprehensionsto the extent that the reader constructs more levels of

    representation and more inferences at each level (Graesser et al 1994: 373). To illustrate

    the multiple levels of representation, consider the following short text:

    The truck driver saw the policeman hold up his hand. The truck driver’s vehicle

    stopped, but a car rear- ended the truck driver.

    The text-base level (i.e. predicate-argument-calculus representation) would

    include a propositional description of the explicit text (Kintsch 1992; Kintsch and van

    Dijk 1978). Accordingly, the first sentence would have the following propositional

    representation:

    Proposition 1: saw (truck driver, Proposition 2)

    Proposition 2: hold-up (policeman, hand)

    Each proposition has a predicate of its own (i.e. verb and/or connective) and one

    or more arguments (i.e. noun or embedded noun-equivalent proposition). The text base

    level would also connect the first / sentence would be connected to the second sentence

    by the overlapping argument “truck driver.”

    Thus, the text- base, providing a shallow representation of the explicit text, also

    computes a referential specification for each noun, while deeper comprehension is

    achieved only with the reader’s constructing motives which explain “why” events and

    actions occurred. Readers would infer that an abrupt stop of the truck caused the car to

    rear end the truck, even though the text never states that there was an abrupt stop! The

    reader would also infer that the truck driver had the goal of stopping the truck and

    performed some intentional action to stop it, even though this was never explicitly

    stated.

    Deeper comprehension stems from the inference-based global message or point,

    of the text, such as” accidents occur even when people follow society’s rules. However,

  • 26

    this level of representation depends on the pragmatic context of the text, such as who

    wrote the text, why it was written, who read the text, and why it was read. Nevertheless,

    according to the inherently relevant inferential components of comprehension, readers

    would be able to generate inferences at all of these levels. Otherwise most readers

    would not / would not be able to construct a detailed description of what the truck driver

    executed to stop the break (i.e. he moved his foot to the brake pedal, he pumped the

    brake, and he calculated the distance between the truck and the policeman).Although

    these details are usually omitted on the text-base level (to avoid wordiness and irrational

    use of experience knowledge), yet it might be important to construct these details when

    an insurance agent is trying to settle an insurance claim for the accident. Thus, the goals

    of the reader and the pragmatic context of the message must also be considered, since it

    is not sufficient to build a theory of comprehension on the basis of the physically

    encoded text alone.

    Some researchers, to cope with the narrow understanding, have enriched the

    definition of comprehension by adopting a “systemic” perspective that appeals to the

    notion of “harmony” (i.e. congruity, compatibility, and synchrony). In one sense,

    harmony addressing the global coherence of the text comprehension succeeds when

    there is unification between both explicit and implicit ideas within the text (Britton and

    Eisenhart 1993). A second sense of harmony addresses the compatibility among the

    three major components of communication system; the author, the text, and the reader

    (Britton and Gulgoz 1991; Rosen blah 1978); Tierney and shanahan 1991). That is,

    comprehension succeeds to the extent that there is harmony among three

    representations; (a) the explicit texts; (b) the author’s intended meaning in the text; and

    (c) the reader’s constructed meaning of the text.

    Writers usually compose the text in Accord with their communication goals,

    whereas readers attempt to recover the writer’s goals during comprehension.

    Comprehension breaks down when there is a gap among the meaning suggested by the

    author, the explicit text, and the reader’s generated meaning and thus it fails to drive a

    transmission of information from author to reader properly. When the reader draws

    appropriate inferences, the comprehension succeeds. For this purpose, the reader

    should/can ask relevant questions to tap potential knowledge gaps, anomalies, and

  • 27

    contradictions. When the reader’s answers to such questions are correct and

    informative, he can paraphrase the message and generate valid inferences. In fact,

    inference generation, question asking, question answering and paraphrasing have

    traditionally been an indicative test of whether readers can understand text without

    using inferential procedures (Kass 1992; Lehnert, Dyer, Johnson, Young, and Harley

    1983; Schank and Abolson 1977). Comprehension also succeeds when the reader is able

    to reveal whether an incoming statement in the text involves a contradiction, an

    anomaly, or irrelevancy with respect to the earlier information (Glenberg, Wilkinson

    and Epstein 1982; Otero and Kintsch 1992; Graesser and Mcmahen 1993; Markman

    1979).

  • 28

    CHAPTER TWO

    2. THE TAXONOMY OF INFERENCES

    2.1. GENERAL PRINCIPLES: INFERENCE NECESSITATING

    FACTORS

    The reason giving rise to so much research in psycho-linguistics and discourse

    processing /comprehension on inferences since the 1970s is connected with finding an

    answer to the following question: which inferences should be generated while others

    cannot be neglected. This question was taken in terms of the classification of inferences

    also in accord with their contribution to text comprehension. (Clark 1977; Graesser and

    Kreuz 1993; Graesser, Singer and Trabasso 1994; Magliano and Graesser 1991;

    Nicholas and Trabasso 1981; Reiger 1957; Singer 1988) , but a consensus has hardly

    emerged, e.g. For Noordman and Vonk (2015:39); a common distinction is between

    necessary and elaborative inferences while Muñoz divides inferences into text-based,

    knowledge- based and causal groups (Muñoz 2014:15-2) .

    Inference generation is obligatory if they provide the so-called text coherence.

    Inferences that are not necessary for coherence are called elaborative, optional, or

    embellishing. Two aspects of coherence can be distinguished (Noordman and Vonk:

    2015:39): Inference making, as some scholars state, allows the reader to extract

    information that is not explicitly described in the text (e.g. Baretta et al 2009) ; Cain

    and Oakhill 1999; Cain et al 2001; Van den Broek 1989); as well as comprehension

    monitoring, which helps readers to verify their understanding of what they have read (

    baker 1989; Kinnunen and Vauras 2010; Vorstius et al 2013; Wagoner 1983; Graesser

    et al 1994; Noordman and Vonk 2015); and the updating of information, in order to

    replace outdated information with new information that is more consistent with the

    context (Carrtti, Belacchi, and cornoldi 2010; Carretti, Borella, Cornoldi, and De Beni

    2009; O’Brien et al 2010; radvansky and Copeland 2010).

    Inference making is crucial in establishing the role and hierarchy of these three

    high-level cognitive processes in comprehension and constructing a coherent situation

  • 29

    model because each of them provides the necessary connections in extracting the

    unexpressed ideas of a text. It is just for this reason that most of comprehension models

    appeal to the process of inference generation (Albrecht and O’Brien 1993; Gernsbacher

    1990; 1997; Graesser et al 1994; Kintsch 1988; 1998; Linderholm et al 2004; Magliano

    et al 1999; Myer and O’Brien 1998; Trabasso et al 1989; Van der broek et al 2005;

    Zwaan and Radvansky 1998).In addition to the multiple taxonomies and the brief

    information describing inference making (e.g. Graesser et al 1994; Zwaan and

    Radvansky 1998; Noordman and Vonk 2015; Muñoz 2014) several properties, usually

    identified as the “back bone of inferences”, also deserve careful consideration:

    In text comprehension inferencing has three important functions: to provide coherence,

    integration and causality. Inferences, as some scholars claim (Graesser et al 1994;

    Magliano et al 1999; Zwaan and Radvansky 1998; Muñoz 2014; Noordman and Vonk

    2015), are of crucial importance in providing coherence to the mental representation of

    the text during reading comprehension. Muñoz (2014: 10) also argues that text

    coherence is gained through establishing connection between the presented information

    and both prior parts of the text and / or prior knowledge. In fact, when coherence

    breaks, readers are forced to “repair” their comprehension through appealing to

    inference making procedure.

    For Hewett (2014; 11), inference generation allows us to “dig out” the implicit

    information and integrate it with explicit information from within the text or by

    incorporating world knowledge with textual information (Cain and Oakhill 2007). For

    other scholars, inferences are also used to derive a high-level, understanding of the state

    of affairs implicitly described throughout a discourse (Cain and Oakhill 2008; Gould

    2008).

    According to Gersbacher (1990), at least four sources of inference providing

    coherence improve the establishment of connection in reading comprehension:

    referential coherence relates to the existing connection between something or someone

    that is mentioned, and a previous reference of that thing/person (e.g.: we got some beer

    out of the truck. The beer was warm); temporal coherence secures necessary relatedness

    to the time (e.g. “I arrived of the start time at 7:45 a.m. The marathon was scheduled to

    begin at 8:00… At eight o’clock sharp the starter fired his pistol”.). Locational

  • 30

    coherence maintains consistency in relation to the space (e.g. “Mike and I were standing

    in the hallway near my office. We were enthusiastically discussing some new data. In a

    nearby office, people had difficulty in concentrating”) and causal coherence explains

    why something happens or what will happen next(e.g. “Brown punched George. George

    called the doctor”). Hence, inference making plays a critical role in constructing

    coherent representations of the text. More than that, text information and background

    knowledge must be integrated in order to shed light on the main idea of a text. Thus, the

    degree of such relevant integration also depends on inferential connections (Magliano et

    al 1999; Zwaan, Langston, and Graesser 1995). Interestingly, the construction-

    integration model (Kintsch 1988, 1998) has distinguished two phases: the construction

    phase, which activates both the current text information and the related prior

    knowledge; and the integration phase, which, in turn,

    (1) activates the more semantically interconnected concepts, and

    (2) deactivates the more disconnected ideas. Therefore, the process of properly

    integrating information into the situation model also requires the suppression of

    irrelevant information (Muñoz 2014: 12). Moreover, several studies have highlighted

    the importance of inference making in establishing causality (Mills and Graesser 1994;

    Singer et al 1992; Walsh and Johnson-Laired 2009; Thagard 2004). The relation

    between causality and inference making goes in both directions: causality facilitates the

    generation of inferences, and inferences help to establish causal relationship - the causal

    network model of comprehension (Trabasso et al 1989; Roberts and Aitken PG no.3).

    Focusing mainly on the role of causal/abductive relationships, bind different types of

    information in a text. As Muñoz states in (2014: 13), readers must be involved in

    effortful and deep reading comprehension in order to generate inferences. Only in this

    way, readers can draw proper inferences (McNamara and Magliano 2009).

    However, inference making is likely to depend on different levels of processing

    going from low demand to high demand of resources (Perfetti et al 2005).In this sense,

    an example of an easily-made inference could be an anaphoric resolution, e.g., in the

    sentences “Ann predicted that Pam would lose the track race, but she came in first very

    easily”, the pronoun “she” is easily mapped into the antecedent “Pam” instead of

    ‘’Ann’’.(Corbetl and Chang 1983); on the other hand, a high demanding inference could

  • 31

    be caused by a causal relationship between information that is poorly connected, e, g., in

    the sentences;’’Joey went to a neighbor’s house to play. The next day, Joey’s body was

    covered in bruises “(Keenan et al 1984 Consequently, inferences may be represented in

    a continue of processing depending on the resources that are necessary to generate

    them:

    1. Direction of processing. Inferences can be driven (as already mentioned earlier)

    by ‘bottom-up “or” top-Bottom/down” processes. Despite the existing disagreement

    between comprehension models to determine the direction by which comprehension

    occurs (Albrech and Myers 1995; Graesser et al 1994; Magliano and Radvansky 2001;

    Myers and O’Brien 1998; Kispal 2008), both text-driven (bottom-up) and goal-directed

    (top-down) processes directly participate in inference making. For example, when the

    text primes reader’s world knowledge without difficulty (“The monkey received a piece

    of fruit as treat”), readers generate automatic inference associated with bottom-up

    processes (e.g. idea of banana’). In contrast, when readers do not have sufficient prior

    knowledge and the text is difficult or unfamiliar (e.g., “Neutrinos are created as a result

    of certain types of radioactive decay, or nuclear reactions such as those that take place

    in the sun, in nuclear reactors, or when cosmic rays hit atoms”), they generate more

    controlled inferences associated with top-down processes, although this does not

    necessarily demand resource-heavy processing. Therefore, the direction of inference

    making in text comprehension, depends on the characteristics of the text (Muñoz

    2014:14).

    2. Knowledge-based information: Coherence integration, sometimes established by

    connecting textual information, requires activating knowledge not present in the text

    (prior knowledge). Explaining information expressed implicitly or in the text make

    readers trigger inferences that go beyond text information; accordingly, the importance

    of inferences in the interaction between text coherence and prior knowledge is of no

    doubt. As some scholars (Mcnamara et al 1996; Azuru et al 2009) state, readers with

    low prior knowledge achieve better comprehension when the text is highly coherent,

    whilst readers with high prior knowledge only benefit from their knowledge when the

    text is low in coherence. But in both cases, accessibility of prior knowledge is of great

    necessity to the reader in order to elaborate inferences (Perfetti and Adlof 2012), which,

  • 32

    in turn, means: inference making depends on memory processes. According to this

    assumption, the Resonance model proposes that current text information and its

    comprehension reactivate prior parts of the text and or knowledge from long-term

    memory (Albrecht and O’Brien 1993: Myers and O’Brien 1998; Myers et al 1994). As

    some theories assume, the existence of a long-term working memory helps text

    comprehension(Ericsson 1995).What it means is that the mental representation of a text

    requires inference making by means of activating memory processes and prior

    knowledge.

    and generic knowledge that are relevant to the text. The specific knowledge For

    Graesser et al (1994; 374), the background knowledge involves both specific includes

    memory representations from other texts, and previous excerpts within the same text

    while the generic knowledge structures include schemata (Mandler 1984; Rumelhart

    and Orton 1977), scripts (Bower et al 1979; Schank and Abelson 1977) frames (Minsky

    1975) , stereo types (Wyer and Gordon 1984), and other structured packets of generic

    knowledge. Most background knowledge coming from life experience, or learning

    organized by hierarchic relations (e.g., a script of eating at a restaurant), are meaningful

    and contextually rich. These rich structures, activated, furnish much of the content

    needed to interpret, explain, predict, and understand narrative events. However, in

    different genres, the background and knowledge structures can be rather abstract and

    decontextualized, such as the Schema for the rhetorical format of a fairy tale (Mandler

    1984; Stein and Glenn 1979). E.g. a novel inference, being a product of several

    cognitive cycles of searching memory, accumulates information from multiple sources

    (Just and Carpenter 1992). The precise mechanisms of constructing these novel

    knowledge-based inferences are not well understood, although several researchers have

    offered speculations (Graesser and Clark 1985; Graesser et al 1994; Johnson- laird

    1983; Schank and Abelson 1977; Wilensky 1983). These novel knowledge-based

    inferences place more burden on the working memory and a potential inference has a

    lower likelihood of being generated on-line to the extent that its generation imposes

    greater working memory (Graesser et al 1994: 374).

  • 33

    2.2. THE TAXONOMY OF INFERENCE

    Much research has been done concerning the analysis of types of the inference

    that exist in their author’s view of comprehension. The literature has been prolific in

    distinguishing various types and categories of inference, ranging from thirteen,

    described in Graesser et al (1994), nine in Pressley and Afflerbach (1995), to the more

    usual two, adopted by many more researchers. Even amongst those scholars who have

    identified essentially the same single distinction between two types of inference, there is

    an assortment of labelling. Commenting on this variety in the naming (and

    characterizing) of inferences, Graesser et al (1994: 374) concluded: Researchers in

    psycholinguistics and discourse processing have proposed several taxonomies of

    inferences... but a consensus has hardly emerged” e.g., “forward” and “backward”

    inferences (Mckoon and Ratcliff 1989; Singer and Ferreira 1983; Van den Broek 1990).

    For van den Broek (1999: 88), the forward inferences are less frequent than backward

    inferences due to the latter’s playing a more essential role in coherence-building.

    Forward inferences (in fact limiting themselves with the cohesive relations based on the

    so-called intralinguistic relational references) are drawn but only when the preceding

    text provides compelling semantic constraints (Klin and Myer 1993;Van den Broek and

    Huang 1995; Vonk and Noordman 1990) .

    The landscape model severely restricts the activity of forward inferences, the

    generation of which depends not only on the presence but also the timing of adequate

    constraints. On the one hand, constraints take time to accumulate: Retrieval from

    episodic memory as well as activation of background knowledge is a slow process

    (Balota and lorch 1986; Bloom et al 1990; Kintsch 1988; Till et al 1988). On the other

    hand, constraints decline when the supporting activation position / Vector dissipates, for

    example, as the time since the last reading cycle increases, this opportunity enables all

    the required constraints allow the forward inference (