Two ways of sticking together and getting along in discourse: propositional cohesion and evaluative cohesion.

In this paper, we intend to describe two systems of what is often referred to as ‘standard’ cohesion, namely entity/propositional cohesion and evaluative cohesion, the first of which has been far more extensively analysed in the discipline of linguistics, especially, grammar, than the latter. Cohesion means, of course, ‘sticking together’. According to Thompson (1996:147-??) cohesion teaches a ‘set of resources’, which ‘the speaker [writer] attempts to employ to enable the listener [reader] to make sense of a piece of communication by ‘organizing the ways in which the meanings are expressed’, by having them connect together in some way. Here we have to underline that the kind of ‘meanings’ held together in standard cohesion practice range from simple entities, objects, people, places to more complex propositions encapsulated in lengthy stretches of text. ‘Standard” cohesion, then affords a set of tools and techniques by which the the speaker [writer] hopes to make the flow of text comprehensible (often named ‘coherent’) to an audience and, in some forms of texts, also engaging. However, the study of standard cohesion can tell us a great deal about how a text is rendered coherent, but it sheds little light on the communicative (the perlocutionary) intents of the speaker [writer] in the first place, that is, why and what it is they wish to communicate and how. A vast amount of human communication involves the expression of evaluation; in essence the appraisal of an entity as good or bad, though good or bad in an infinity of different ways. We very rarely discuss entities or propositions without evaluating them in some way. Indeed the presentation and arrangement of information without the speaker [writer] evaluation would not only by very dry but largely uninformative on an interpersonal level. Texts then are also held together, they cohere, in terms of the evaluations they express, and it is the study of evaluative coherence (sometimes referred to as evaluative harmony) which sheds light on what speakers [writers] intend to do when they communicate to others. As Aristotle noted, human communication largely consists in attempts to connnect with and to influence the beliefs and even behaviour of other people (Partington, Duguid and Taylor 2013 ??), in other words, to persuade them (of everything from the fact that you are a person worth listening to, to how they should spend their money, to how they should vote.) In order to study how evaluative cohesion functions in detail, we will utilise concordancing of relevant lexical items, lexical templates often called units of meaning, as they appear in the Siena-Bologna (SiBol) Modern Diachronic Corpora suite of corpora. This consists of four sister corpora, the first three of UK newspaper texts from different but contemporary periods in time, designed and compiled to be as alike as possible to eliminate potentially complicating variables. They contain all the articles published by the three main UK broadsheet or so-called ‘quality’ newspapers, namely The Times, the Telegraph and the Guardian in the years 1993 (the SiBol 93 corpus), 2005 (the SiBol 05 corpus) and 2010 (the SiBol 2010 corpus). They contain, respectively, circa 100 million words, 150 million and 140 million words. The 2013 corpus wave, instead, contains the output of that year of 12 English language newspapers, including the original The Times, the Telegraph and the Guardian plus two UK tabloids, the Mirror and the Mail, two US newspapers, the New York Times and the Washington Post, the Times of India, the South China Morning Post (Hong Kong), Daily News (Egypt), Gulf News (UAE), This Day Lagos (Nigeria). It contains a total of 327 million words