In this chapter I will sketch work on information structure pre Halliday. This work will include mention of Aristotle, Weil and Prague School work e.g. Mathesius and explain his three part definition of theme. I will also briefly mention the work of Firbas and Daneš (discussed in detail in chapter 3) in order to contextualise what Davidse calls Halliday’s separating theme. The separation of the theme and information systems is a distinguishing feature of SFL and the chapter will describe how the systems fit into the textual metafunction and into the wider SFL architecture.

In this chapter I will detail the Hallidayan model by describing the systems of tonality and tonicity in detail. Particular attention will be paid to three problems which have been raised in the literature. Firstly how long an item which is available in the context remains recoverable. Secondly the informational status of pre-tonic accents. Thirdly the relation between clauses and information units in cases of marked tonality which raises issues as to what is actually meant by information. In IFG Halliday defines New information as the main point of the message but it is rather difficult to see how this applies in cases of marked tonality. A solution based on early work by Halliday the paraphone will be suggested though this will be amended and extended in chapter 4 by introducing the concept of an increment.

In this chapter I will examine other functional and pragmatic theories of information structure in order to see if insights from these theories can help overcome the issues mentioned in the previous chapter. The theories examined will be (i) FSP which examines information structure in terms of communicative dynamism as operationalised by the combined presentation and quality scales set out by Firbas, (ii) work which questions the binarity of Given and New and presents the information status of referents in a tripartite scale e.g. Prince and Kaltenbock. These scholars sees the lexicogrammar as a series of contextualising cues which allow hearers to infer informational status in individual contexts (iii) the functional syntactic model associated with Lambrecht which argues that sentence structure reflects speaker’s assumptions of the hearer’s knowledge and awareness. This leads to an analysis of propositional information such as presuppositions and assertions, identifiable information in relation to activation and referential information in relation to topic. I will take issue though with some of the claims re universal cognitive processes (iv) the givenness hierarchy as proposed by Gundel and colleagues. This hierarchy focuses on a mix of referential and identifiable information and I will argue unnecessary blurs the distinction between the two (v) interactional sociolinguistic approaches to units such as TCUs and (vi) linear incremental grammars which see language use emerging as a sequences of anticipations. Both (v) and (vi) are concerned with propositional information and have much to say about how propositional information is exchanged. In addition there will be a brief critical review of shared knowledge and theory of mind in order to illustrate a mechanism where interactants can be said to share information in the broadest sense. I will argue ultimately against theory of mind while recognising that people are disposed to treat others as if they have access to the same culturally specific knowledge. This is a crucial chapter as it aims to show the state of the art in non SFL theories of information structure in spoken English and to disentangle the different meanings of information structure.

In this chapter I will propose an expanded model based on three different information scale: (i) a referential one based primarily on lexis, (ii) an identification one based primarily on phonology (iii) and a propositional one realised by the combination of lexicogrammar and prosody. Such a model will be able to track the unfolding of expectations as they emerge through the production of lexical items and also to model the relation between propositions. The model will be incorporated into the architecture of SFL and relevant systems drawn.

The model will be tested and revised by examining spoken data. This will be a contextualised and qualitative analysis and while I will employ corpora such as IVIE and London-Lund (hopefully version 2) as well as other extracts of speech from a range of genres I will not employ corpus linguistic methods. This is because I am uninterested in frequency but am rather interested in how speakers in different contexts negotiate their common understandings. To study I need extended extracts accompanied by relevant sound files from different genres. This chapter will show how speakers can if needed manipulate information structure at the scales of (i) identification/activation and (ii) establishing common ground.

I start the chapter by synthesising what information is and the functions that are realised in the expanded model. I will then consider how such functions are realised in languages other than English especially ones where prosody is of no or of limited importance respectively Japanese and Greek. Study of these languages will show that these languages rely more on ellipsis, morphology, case and syntax in projecting Information Structure. I will show that metaphors such as ‘point of departure’ are not helpful in understanding how speakers form wordings into text. The book will conclude with some personal reflections on what I feel are the outstanding issues in the study of information structure within an SFL framework.