This thesis explores the domains of sublanguage, machine translation and textual analysis. Chapter 1 discusses the definitions and characteristics of sublanguage put forward by researchers to date, as well as the background of textual analysis in linguistics. This discussion reveals that, although there is much to be gained from textual analysis, little consideration has been given to the notion of "text" in the sublanguage approach to machine translation (MT). Before any sublanguage analysis can proceed, compilation of a corpus is necessary. To date, attention has been focussed on the criteria for compiling general language corpora. Chapter 2 addresses the problems of compiling corpora for sublanguage research and offers guidelines for this purpose. An exploration of the advantages of considering text type and communicative function in the sublanguage approach to MT is the focus of Chapter 3. Three text types with a similar communicative function from the same highly restricted sublanguage domain are compared for linguistic features which cause semantic, syntactic and lexical ambiguities. Finally, Chapter 4 summarises and evaluates the results obtained in Chapter 3. Conclusions are drawn about "text type" and communicative function and about the advantages of considering "text" for MT.