In this paper, I first present a formalized classification of question sentences and a study on the automatic generation of paraphrases of interrogative sentences. These results can be applied at different levels in NLP for the Quechua language. Interrogative sentences are an important part of communication, since they allow one to obtain information, engage in conversation, or show curiosity. But in Quechua, how many forms can be formed correctly? What are the different types of interrogative sentences in Quechua? For this study, we gathered texts from ancient chronicles, contemporary publications, and transcriptions of oral recordings of native speakers. The analysis of this corpus allowed us to identify some morphosyntactic and semantic patterns and forms of interrogation. Then, using the NooJ platform, we formalized the grammatically valid transformations and developed grammars that can generate paraphrases of a given Quechua interrogative sentence.