This paper will present a translation project with second-year bachelor students together with some observational data. The students firstly translated an academic abstract from Italian (their first language) to English by using the language resources they were acquainted with. Then they translated a second abstract by using a pre-compiled offline corpus of academic articles. The paper findings highlight that the second translations were better as regards grammar and word choices, while sentence structures still showed influences of the students' first language. The students learnt that a corpus is a useful supplement to a dictionary, as a corpus may corroborate what the dictionary suggests. The report also highlights the importance of corpus analysis training in order to help users process corpus information, which otherwise might be perceived as "distracting".