This dissertation presents a computational interpretation of futuro do pretérito. It is integrated to Rodrigues' model [Rod95], which covers three tenses of Portuguese: pretérito imperfeito, pretérito perfeito and pretérito mais-que-perfeito. These tenses are similar to English past tenses. On the other hand, futuro do pretérito is similar to "future in the past" sometimes denoted by "would" or "was going to". The introduction of futuro do pretérito makes the model more complete, once this tense is used in narrative texts to describe facts occurred after facts narrated in phrases using the other tenses in question.
The developed interpretations are based on grammars [CC01, FM95] and on linguistic studies [Coroa85, Ilari97]. Nevertheless, these sources have been insufficient for modelling "futuro do pretérito" interpretation in contexts of discourses, once these contexts reduce or change "expected behaviour" of tense. Our work analyzes futuro do pretérito considering the discourse context.
Some of most important formalisms for temporal reasoning and interpretation, namely the situation calculus [McCHay69], the event calculus [SK95] and the interval temporal logic [All84], are analyzed and evaluated in terms of their adequacy to treat information got from interpretation of phrases in natural language.
Both, Rodrigues' work on tense interpretation [Rod95] and Rodrigues and Lopes framework for temporal representation [RL92, RL92b, Rod95] are discussed. Some changes are made in order to include futuro do pretérito in the framework.
The rules of the prototype are introduced and implemented in Prolog extending the original Rodrigues' implementation, which is also discussed.