challenge, fend off or restrict the scope of the alternative positions and voices” (Martin and White 2005: 102). That is to say, the writer supports only one of these positions. Expressions of Contraction are used when the writer feels the need to lay emphasis on commitment to the truth of the information that s/he is transmitting because it is not obvious to others and hence challengeable. Therefore, Contraction seems out of place when there is no room for challenge: for example, certainly would be odd in I’m certainly wearing a green coat in a face-to-face conversation. Contraction is divided into two subtypes: Proclaim and Disclaim.
The subcategories of Proclaim are Concur, Pronounce and Endorse. Concur “involves formulations which overtly announce the speaker / writer as agreeing with, or having the same knowledge as, some projected dialogic partner”. (Martin and White 2005: 122). That is, the statement is presented as agreeing with, or having the potential to agree with, the majority of voices. Expressions of Concur include adverbials that express agreement with previous expectations (of course, naturally, not surprisingly, admittedly…), strong evidential expressions such as clear, evident, obvious and the derived adverbs, and other ways of indicating agreement with other opinions (as everyone knows, it is well-known that, it is acknowledged that, no-one would deny that…), as well as the Spanish equivalents of these expressions. Concur also includes those expository questions that assume an obvious answer (Martin and White 2005: 123), such as (15) and its Spanish translation (16):
(15)Does anyone in their right mind think that any country would willingly put itself through what Greece has gone through, just to get a free ride from its creditors? (EO_ESS_001)
(16)Alguien en su sano juicio cree que algún país estará dispuesto a atravesar voluntariamente lo que Grecia ha tenido que atravesar, sólo por conseguir ventajas de sus acreedores? (STrans_ ESS_001)
The second subtype of Proclaim, Pronounce, “covers formulations which involve authorial emphases or explicit authorial interventions or interpolations” (Martin and White 2005: 127). That is, the author expresses that his/her opinion is firm, without referring to other opinions. Realisations of Pronounce include emphatic affirmation, expressions of epistemic certainty (certainly, definitely, really, surely, for sure…), lexical verbs referring to speech acts or mental states of certainty, in the first person (I know, I say…), other expressions which insist that the facts are real (the fact is that…), and even parallelisms or repetition of words.
The third subtype, Endorse, “refer[s] to those formulations by which propositions sourced to external sources are construed by the authorial voice as correct, valid or undeniable or otherwise maximally warrantable” (Martin and White 2005: 126). Endorse resembles Attribute in that an external source is mentioned, but in this case the writer supports only the position expressed by the source, thus expressing high commitment to the information transmitted. Expressions of Endorse include verbs such as show, prove, demonstrate, find or point out with different persons from the first.
The Disclaim category challenges some contrary position, by openly rejecting it or by positioning itself at odds with it. Its subcategories are Deny and Counter. Deny consists in the overt negation of a proposition. Cases in which negation affects only part of the clause, as in (17), have also been included, since the writer’s intention is still to reject the idea that Greece should bear the consequences. Verbs with negative meaning such as lack, fail or neglect have also been considered as cases of Deny, following Mora (2011: 65).
(17)If Europe has allowed these debts to move from the private sector to the public sector – a well-established pattern over the past half century – it is Europe, not Greece, that should bear the consequences. (EO_ESS_001)
The other subcategory, Counter, “includes formulations which represent the current proposition as replacing or supplanting, and thereby ‘countering’, a proposition which would have been expected in its place” (Martin and White 2005: 120). In short, this category concerns counter-expectation. Among the many realisations of Counter, the most frequent are conjunctions and connectives of contrast such as although, however, yet, but, adverbials such as even, only, just, still, already or yet, and the Spanish equivalents of all these expressions. Counter also includes the adverbials actually and in fact and Spanish correlates such as en realidad or de hecho.
3 The Research Hypotheses
Before embarking on the actual Engagement analysis of the texts, three hypotheses were set. The first was that expressions of Engagement tend to be faithfully translated; in order to check this hypothesis, the texts analysed (see Section 4) include English and Spanish originals and their translations. A corollary of the first hypothesis was the second hypothesis, namely that a comparison including only the original texts in both languages would show greater differences: the English and Spanish essays would tend to favour the use of language-specific preferred devices. The third hypothesis was that the distribution of Engagement expressions would differ depending on text type: Contraction expressions could well be more numerous in the argumentative texts, since the writer has a greater need than in expository texts to defend his/her position against other possible alternatives; by contrast, Expansion devices might well be more common in the expository texts to signal limitations of the present state of knowledge. These hypotheses will be tested by comparing the data in the ways specified in Section 4.
4 The Data, Software and Method
The texts analysed here were extracted from the MULTINOT corpus, which consists of original and translated texts in both directions and is designed as a multifunctional resource to be used in different disciplines, such as corpus-based contrastive linguistics, translation studies, machine translation, computer-assisted translation and terminology extraction. The MULTINOT corpus, described in more detail in Lavid et al. (2015), includes a wide range of registers from the written mode, following typologies used in other parallel corpus projects, such as the DPC corpus (Paulussen et al. 2013) and the CroCo Corpus (Hansen-Schirra et al. 2012). The registers included are the following: novels and short stories; news reporting articles; manuals and legal documents from webpages; official speeches and proceedings of parliamentary debates; annual reports and letters of self-presentation of companies, promotion and advertising brochures; scientific texts; essays; and popular science expository texts.
The data chosen for analysis were 40 texts belonging to the last two categories: 20 essays and 20 popular science expository texts. The essays were written in the 2000’s, and the expository texts from 1980 onwards. The texts were also evenly divided according to the criteria of language (20 English and 20 Spanish) and originality (20 are original texts, and 20 are their translations).
The references and URLs of all the texts are specified in the Appendix. The argumentative texts are political essays on economics; the English originals were extracted from the non-profit international organisation Project Syndicate, which publishes and syndicates opinion articles on topics such as global affairs, economics, finance and development, and has members in many countries around the world. This organisation also provides the Spanish translations. The Spanish argumentative originals were extracted from the quality newspaper El País, and their translations were downloaded from the URL ‘Essay and science’, a webpage aimed at spreading original essays written in Spanish. The English and Spanish argumentative essays sometimes include short biographical notes about the authors. These parts were excluded from the analysis, since they lie outside the texts proper. The English and Spanish expository texts were extracted from highquality books by prestigious authors and publishers, aimed at the dissemination of knowledge in the areas of science and social science.
Many of the texts contain 1,000 words approximately; the others were cut after the paragraph to which the 1000th word belonged, so as to maintain a balanced number of words. Therefore, the 40 texts analysed amount to approximately 40,000 words.
The quantitative analysis was carried out with the aid of the UAM Corpus Tool, a free tool created and regularly updated by Mick O’Donnell at the Universidad Autónoma de Madrid.4