This paper summarizes the results of our first participation at ResPubliQA. Lacking a true question answering system, we relied on our traditional search engine to find the answers. We submitted two runs: a pure IR baseline and one where definition question identification and corpus-specific techniques were employed to improve IR performance. Both runs performed well, and our second run obtained better results than the baseline. However, the drawbacks of relying solely on an IR phase are clearly seen in our runs.