In this chapter, we present a corpus-based perceptual study of the French /y/ and /u/ produced by Spanish learners of French, as carried out in the framework of the InterPhonologie du Français Contemporain (IPFC) project. In order to evaluate the quality of realization of our learners' productions, we performed an experiment in which French native listeners had to identify the vowel (/y/, /u/) of four monosyllables produced by two groups of Spanish learners (one from Geneva and one from Madrid) and a group of native French speakers in two tasks (repetition and reading). The results revealed a better identification rate for /u/ than for /y/, suggesting that a new vowel, such as a vowel that has no phonemic and phonetic equivalent in L1 (/y/), is more difficult to learn than a vowel that is phonemically but not phonetically similar (/u/). Also, for /y/, orthography seems to interfere negatively with more errors for the words produced in the reading task. On the contrary, orthography acts positively for /u/, with better performance for the words produced in the repetition task. Lastly, although our results surprisingly showed better performance for the Spanish learners in Madrid, a detailed analysis revealed that this difference has to be attributed to individual differences among learners.