The author adopts the text-based framework in semantic inferences and develops a computational linguistic model for identifying and decomposing Textual Entailment. The model is developed according to applications criteria i.e. RTE-1-4 pairs taken from various application domains; RTE-5-6 for summarization needs; AVE7 etc. The aim was to shed light to a) linguistic phenomena and devices underlying textual inferences and b) the way the latter interact with different types of inferences. Semantic and inferential phenomena such as deductive versus inductive arguments, quantifiers, plurals, anaphora and ellipsis are examined. The study shows that the decomposition methodology proposed can be applied on text entailment.