THEME AND FOCUS: CROSS-LANGUAGE COMPARISON VIA TRANSLATIONS FROM EXTENDED DISCOURSE. (VOLUMES I AND II) (FUNCTIONALISM, TRAVNICEK, BRAZILIAN PORTUGUESE, SYSTEMIC LINGUISTICS, HALLIDAY)

The concept of theme/rheme structure is frequently conflated with that of given/new information, theme being defined as "that which is known or at least obvious in the given situation, and from which the speaker proceeds in his discourse" (Mathesius 1939 1947 :234). This view fails to account for themes that are not "given" and for given information that is not in initial position. With a systemic approach, on the other hand, theme and information are independently variable. In the model of Halliday (1967-68), the systems of theme and information together make up the resources that are essential to the creation of text.;Examination of the corpus revealed consistent patterns of both theme and focus. Focus was upheld in translation in 87 percent of the working units; the themes, which were often multiple, showed greater variation.;A message-based model is proposed in which three systems provide the resources that are essential to contextualization: theme, information, and message-framing. The theme specifies the relationship between the speaker's thought (Travnicek 1962) and the onset of a message. It is always expressed through the initial constituent(s), for which there are three order classes: conjunction, thematic adjunct, and obligatory major theme. Left-to-right progression is generally from noncognitive to cognitive. The information system provides the mechanism by which the listener can compute an antecedent and thereby integrate new information into his data base. Messages are delimited by onset and closure, both of which are essential to segmentation.;These systems come to the fore in translation, which must capture all the meaningful components of a text and re-create them in another language. It was hypothesized that the structural realizations of theme and information, which have written reflexes, are carried over in translation except in the presence of constraints. A 54,040-word corpus of 32 Brazilian Portuguese texts was examined against matching published translations in English. Questionnaires based on the corpus were administered to two study populations, 22 subjects who knew Portuguese (including translators) and 26 who only knew English.